I want to start with a confession. I have been pro crypto for years, and normally that admission alone gets me dismissed by the very analysts and portfolio managers who read this newsletter. So instead of asking you to trust my conviction, I am going to do something different this week. I am going to argue against myself first. I pulled the bear case straight from Seeking Alpha, checked the numbers on Yahoo Finance and Macrotrends, and only then built the trade. If you still think crypto is a casino after reading this, fine. But I don’t think you’ll think that by the end.

What “asymmetric” actually means, in plain English

Before the data, a quick beginner detour, because this word gets thrown around a lot without being explained.

An asymmetric bet is simply one where the potential loss and the potential gain are not the same size. If you can lose 50% but make 300%, that is asymmetric. Most professional investing is actually a search for this exact shape. Venture capital is built on it. Options traders live and breathe it. And right now, Bitcoin is sitting in one of the more extreme asymmetric setups it has offered in years, not because the price is guaranteed to rise, but because the size of the plausible downside has shrunk while the size of the plausible upside has not.

That distinction matters more than any single price prediction I could give you.

The setup nobody wants to admit is interesting

Bitcoin is currently trading in the low $60,000s, down roughly half from its all time high above $126,000 set in October 2025. The Fear and Greed Index sits around 23, in Extreme Fear territory. Bitcoin’s Sharpe ratio, which measures return per unit of risk, has fallen to its lowest reading since 2022, meaning that on a purely risk adjusted basis, holding Treasuries would have beaten holding Bitcoin recently.

That sentence alone should tell you why most people are staying away. Nobody wants to catch a falling knife, and the data confirms the knife has been falling.

But here is the part that gets buried under the headlines. Every prior Bitcoin bear market has followed a shrinking pattern of drawdowns. The peak to trough declines across cycles have gone from roughly 94%, to 87%, to 84%, to 77%, and this cycle is tracking closer to the 50 to 70% range. Each cycle, the market has matured, the drawdown has gotten shallower, and the recovery has arrived faster. That is not a guarantee this time follows the same script. It is a pattern, not a law of physics. But it is a pattern with a fifteen year sample size, and dismissing it entirely is its own kind of bias.

Steelmanning the bears, because I promised I would

I searched Seeking Alpha specifically for the strongest bear arguments, and I am not going to sand down the edges.

One analyst, who discloses an active short position in Bitcoin, argues the current cycle mirrors past liquidity driven selloffs and expects a bounce toward resistance to ultimately fail, with downside support sitting near $62,500. Another bear case lays out a specific target of roughly $36,000 by late 2026, framing this as a “boring” multi quarter grind lower rather than a sharp crash, on the view that this is simply the bear phase of the four year halving cycle playing out on schedule.

A separate piece on a Bitcoin mining and treasury company puts numbers behind the pain for the industry that lives and dies by the coin price, rating that stock a strong sell as it trades above net asset value while facing ongoing declines.

I take these seriously. Mining economics genuinely do get squeezed in drawdowns. Corporate treasury buyers genuinely can turn from marginal buyer to marginal seller. Strategy, the largest corporate holder of Bitcoin, sold roughly $216 million worth of coin in early July just to fund dividend payments on its preferred stock, and it was not the first such sale in recent weeks. That is a real, structural source of supply hitting the market from a name that used to be the poster child of the “never sell” thesis.

None of that is bullish in isolation. But asymmetry is not about ignoring the bear case. It is about weighing it against what the other side of the trade offers, and here is where it gets interesting.

Why the downside looks capped and the upside doesn’t

On chain data just flipped a signal that has coincided with every major cycle bottom in Bitcoin’s history: the amount of Bitcoin supply sitting at an unrealized loss recently overtook the supply sitting in profit, with roughly 10.5 million coins underwater against 9.8 million in profit. Historically, this crossover has marked the depths of despair rather than the start of a new leg down. Bitcoin also touched its 200 week moving average, a long-term support level that has held in every previous bear market to date.

Meanwhile, institutional selling has not been what the headlines imply. A record run of Bitcoin ETF outflows made for scary headlines, but the composition of that selling tells a different story. Hedge funds and brokerages, the fast money cohort, cut their ETF positions by 39% and 53% respectively. Investment advisers, the largest holder cohort by size, trimmed their positions by just 5.9%. Fast money left. Allocation money mostly stayed. That is not what capitulation from long-term holders looks like.

Layer on the institutional price targets still on the table. Standard Chartered, despite cutting its target, still sees $100,000 to $150,000. JPMorgan’s gold parity framework puts fair value near $170,000. Fundstrat’s Tom Lee is at $150,000 to $250,000. Even the most bearish widely cited institutional view, Bloomberg Intelligence’s mean reversion scenario, is a tail case rather than a base case.

Put the ranges side by side. The believable bear case clusters around $36,000 to $55,000, a further decline of roughly 15 to 40% from current levels. The believable bull case clusters around $100,000 to $250,000, a gain of roughly 60 to 300% from current levels. That is not a coin flip. That is a lopsided bet, and lopsided bets are exactly what this newsletter exists to find, across any asset class.

How I am actually sizing this, cross asset

This is the part I think gets skipped in most crypto commentary, and it is the part that matters most to the professionals reading this.

Asymmetric does not mean “go all in.” It means size the position so that being wrong does not hurt you, while being right changes your year. Think of it the way an options trader thinks about a long-dated call. You are not betting the house. You are buying a ticket to a distribution of outcomes that is skewed in your favor, and you are sizing it so the premium, in this case the drawdown risk, is something you can shrug off.

There is also a cross asset angle worth flagging for the macro thinkers in the room. Bitcoin’s setup here rhymes with a broader theme we have covered before: assets priced for permanent bad news tend to offer the best risk adjusted entries, whether that is emerging market equities after a currency crisis or commodity producers after a demand shock. Bitcoin, right now, is priced like the bear case is certain. It isn’t. It is one plausible outcome among several, and it is not even the consensus one among institutions with the largest checkbooks.

What would actually kill this thesis

I want to be honest about what would prove me wrong, because a real thesis needs a real falsification point. If Bitcoin closes a week below $59,000, the credible bear case opens up toward the mid $50,000s and, in a genuinely ugly scenario echoing 2022, into the mid $40,000s. If investment advisers start liquidating at the same pace as hedge funds did, the “allocation money stayed” argument collapses, and with it a big chunk of my conviction. Watch that cohort, not the headline outflow number.

That is the risk. I am not hiding it. But it is a bounded, identifiable risk against an upside case with no natural ceiling, backed by institutions that are, in aggregate, still buyers.

The trade

I am not telling you to mortgage your house. I am telling you that when extreme fear, a shrinking historical drawdown pattern, a capitulation signal on chain, and a bifurcated seller base all show up in the same month, that is not noise. That is the setup this newsletter was built to find.

You will never have to ask what the trade is. This week, it is sizing into weakness while the crowd is still asking whether crypto is dead.

This is not financial advice. Position sizing and risk tolerance are personal. Do your own research before acting on anything in this newsletter.

Upgrade to Paid

Most retail traders start with the same toolkit: a handful of indicators, a couple of trendlines, and a gut feeling. It’s easy to learn and it feels like trading.

But there’s a completely different world of market participants — banks, hedge funds, prop firms — who read price action through a different lens entirely: order flow, volume, and positioning.

This is where real technical analysis lives, and it’s a different animal from the RSI-crosses-70 strategies that flood social media.

This is a long, detailed walkthrough. It covers why price behaves the way it does at a mechanical level, the core technical analysis tools professional swing traders actually rely on, how to read volume and market structure like an auction, how to layer macro and positioning data on top of your charts, and how to build a repeatable entry and risk framework around all of it.

Table of Contents

1. Why Price Moves the Way It Does

2. The Building Blocks of Price Action

3. Support, Resistance, and Market Structure

4. Volume: The Missing Half of Every Chart

5. Auction Market Theory and the Volume Profile

6. Classic Indicators — What Actually Still Works

7. Reading the Macro Backdrop

8. What Drives Each Asset Class

9. Following Institutional Positioning

10. The Entry Models

11. Risk Management and Position Sizing

12. Building a Trading Plan

13. Common Mistakes and How to Avoid Them

14. Final Thoughts

1. Why Price Moves the Way It Does

Every market runs on an order book — the live menu of buy and sell offers waiting to be filled. A retail trader with a modest account can buy or sell “at market” and barely move the price.

The cost of that convenience, the bid-ask spread, is usually a few cents or a few dollars.

An institution trying to deploy hundreds of millions of dollars faces a completely different problem.

There simply isn’t enough resting liquidity in the book to fill an order that size at one price. Buying it all at once would mean sweeping through the book at increasingly worse prices — and unwinding the position shortly after could mean losing a staggering amount to slippage.

So large players don’t dump size all at once. They break enormous orders into thousands of smaller ones and feed them into the market gradually, sometimes over days or weeks, using execution algorithms designed to minimize cost.

This single fact is the foundation of technical analysis as professionals use it: because filling and later unwinding a massive position takes time, institutional flow creates persistent, gradual drift in price, and that drift leaves visible fingerprints on the chart — in volume, in the shape of consolidations, and in how price reacts around certain levels.

Upgrade to Paid

Academics call the resulting phenomenon order-flow autocorrelation — a run of buy orders tends to be followed by more buy orders.

Technical analysts call the visible result a “trend.” Understanding that a trend is really just the visible residue of a large, slow-moving order changes how you read every chart you’ll ever look at.

2. The Building Blocks of Price Action

Before layering on volume, auction theory, or macro context, every technical trader needs fluency in the raw language of price: the candlestick.

Candlestick anatomy. Each candle represents a battle between buyers and sellers over a fixed period — one minute, one hour, one day.

The body shows the range between open and close; the wicks (or shadows) show the full high-to-low range.

A long body with small wicks shows conviction in one direction. A small body with long wicks — a doji or a spinning top — shows indecision, often appearing right before a shift in control.

Common candlestick patterns worth knowing:

• Engulfing candles — a candle whose body completely swallows the previous candle’s body, often marking a short-term reversal in momentum.

• Pin bars / hammers and shooting stars — a long wick rejecting a price level, showing that aggressive buyers or sellers stepped in and were overwhelmed.

• Inside bars — a candle fully contained within the previous candle’s range, showing a pause or coiling before the next expansion.

• Doji clusters — multiple small-bodied candles in a row, usually marking indecision at a key level before a breakout or reversal.

  Upgrade to Paid

Timeframe alignment. Professional swing traders rarely trade off a single timeframe. A common approach is a three-tier system: a higher timeframe (weekly or daily) to establish the dominant trend and macro bias, a medium timeframe (4-hour or daily) to map out structure and key levels, and a lower timeframe (1-hour or 15-minute) purely for entry timing.

Trading against the higher timeframe trend is possible but statistically harder — you’re fighting the direction the “big money” order is most likely still filling in.

3. Support, Resistance, and Market Structure

Classic support and resistance is often oversimplified as “a line where price bounced once.” Institutional-grade technical analysis treats these levels as zones of concentrated past transaction activity — places where a large number of orders were filled, meaning a large number of traders have a vested interest in defending or reacting to that price again.

How to actually map structure:

• Swing highs and swing lows define the skeleton of a trend. A sequence of higher highs and higher lows defines an uptrend; lower highs and lower lows define a downtrend. The moment that sequence breaks — a lower high after a series of higher highs, for example — is the first technical warning sign that the drift may be shifting.

• Zones over lines. Instead of drawing a single line, mark a zone spanning the wicks and bodies of the candles that reacted at that level. Price rarely respects a level to the exact tick; it respects an area.

• Role reversal. A broken resistance level frequently becomes new support (and vice versa), because the traders who were trapped or hesitant at that level now have a new reference point to defend.

  Upgrade to Paid

4. Volume: The Missing Half of Every Chart

Price tells you what happened. Volume tells you how much conviction was behind it. A breakout on razor-thin volume is far more likely to fail than a breakout accompanied by a surge in participation, because thin volume means few real participants have committed capital to that move.

Key volume signals:

• Volume expansion on breakout — confirms genuine institutional participation.

• Volume divergence — price makes a new high, but volume shrinks compared to the prior high. This often precedes exhaustion.

• Climactic volume — an extreme volume spike, often at the end of a strong trend, can mark capitulation (in a downtrend) or blow-off euphoria (in an uptrend) — both classic reversal signatures.

• Low-volume pullbacks — in a healthy trend, corrective moves against the trend typically occur on lower volume than the impulsive moves in the trend’s direction.

5. Auction Market Theory and the Volume Profile

This is where technical analysis stops being about lines and starts being about where the money actually transacted.

Auction Market Theory treats every market as a continuous double auction between buyers and sellers. Prices spend most of their time in balance — a range where buying and selling pressure roughly agree on fair value, generating heavy trading activity. Plot the volume traded at each price level over a session, and you get a volume profile: a horizontal histogram overlaid on the price axis.

Key volume profile concepts:

• Point of Control (POC) — the single price level with the most volume traded. This is the “fairest” price in the range and tends to act as a magnet.

• Value Area (VA) — typically the range containing roughly 70% of the volume around the POC. This is the fair value zone.

• Value Area High / Low (VAH / VAL) — the upper and lower boundaries of that fair value zone; these behave like dynamic support and resistance.

• Low-volume nodes (LVNs) — thin areas on the profile where very little trading occurred. Price tends to move through these areas quickly, because few traders have a reason to defend them.

• High-volume nodes (HVNs) — thick areas where price tends to slow down, consolidate, or reverse.

  Upgrade to Paid

At some point, an imbalance in order flow — new information, a shift in institutional appetite, a macro catalyst — breaks the balance and pushes price into price discovery, searching for a new fair value.

This is genuinely a liquidity-seeking auction: aggressive buyers keep paying higher and higher prices because they’re more interested in getting filled than in getting a good price, exactly like bidders at an art auction chasing a painting. Eventually sellers agree the new level is fair, trading slows, and a new value area builds — the market has re-entered balance at a new location.

Reading a chart through this lens means constantly asking: is the market building value (balance, low volatility, thick volume profile) or discovering value (imbalance, expansion, thin volume profile)? That single question does more work than most indicator combinations.

6. Classic Indicators — What Actually Still Works

Indicators aren’t magic, but used as confirmation tools layered on top of structure and volume, several remain genuinely useful:

• Moving averages (50/100/200-period) — best used as dynamic trend filters rather than standalone signals. Price consistently holding above a rising 50-period moving average on the daily timeframe is a simple, robust trend filter.

• RSI (Relative Strength Index) — most useful not for “overbought/oversold” but for spotting divergence: price makes a new high while RSI makes a lower high, warning that momentum is fading even though price hasn’t turned yet.

• MACD — useful for gauging the rate of change of momentum, particularly the histogram, which often peaks before price does.

• ATR (Average True Range) — not a directional tool at all, but essential for position sizing and setting stop-losses that respect the market’s actual volatility rather than an arbitrary percentage.

• VWAP (Volume-Weighted Average Price) — essentially an intraday volume profile POC, heavily used by institutional execution desks as a benchmark; price reactions around VWAP are common on lower timeframes.

The mistake most beginners make is stacking five indicators that are all mathematically derived from the same input (price), producing five versions of the same signal and an illusion of confirmation. A better approach: one trend filter, one momentum/divergence tool, one volatility tool, and the rest of the picture built from structure and volume.

Upgrade to Paid

7. Reading the Macro Backdrop

Technical structure tells you where institutions are likely reacting. Macro tells you which direction they’re structurally biased to trade. The two work together — macro sets the weather, technical analysis picks the exact door to walk through.

Every major central bank operates under some version of a dual mandate: stable prices (inflation, typically targeted near 2%) and maximum employment.

When inflation runs hot, the central bank hikes rates to cool borrowing and spending. When unemployment rises, it cuts rates to stimulate credit and spending.

Reading incoming data — CPI, PCE, and PPI for inflation; non-farm payrolls, the unemployment rate, JOLTS, and jobless claims for the labor market — lets traders anticipate which direction policy is likely to move before the announcement, because markets tend to price in expectations well ahead of the actual decision.

Because policy shifts are slow and gradual, they produce exactly the kind of multi-week, multi-month drift that technical structure and volume profiles are built to capture.

A rising-rate environment doesn’t just “feel” bearish for gold — it shows up on the chart as expanding value areas trending lower, breakdowns through low-volume nodes, and repeated rejections at Value Area High.

Upgrade to Paid

8. What Drives Each Asset Class

Bonds

A bond is essentially an IOU: lend money to a government or company, get it back later with interest — the yield. That yield is built from the central bank’s base rate, a time premium, a risk/credit premium, and an inflation premium. Plotting yield against maturity produces the yield curve, which normally slopes upward.

When short-term yields spike above long-term yields — an inverted yield curve — it has historically preceded nearly every U.S. recession, making it one of the most reliable macro-technical signals available.

Forex

Currency pairs are frequently traded on technicals alone, but the underlying driver is comparative bond yields. When one country’s central bank turns more hawkish (or less dovish) relative to another, the yield differential shifts, and that shift tends to produce the kind of clean, sustained trend that technical structure captures well — clear breaks of value, retests, and continuation.

Gold

Gold’s core role is as an inflation hedge and safe haven, but it competes directly with bonds for the same pool of capital. The deciding factor is the real yield — the nominal bond yield minus expected inflation. Falling real yields tend to coincide with gold breaking into new, higher value areas; rising real yields tend to coincide with gold stalling or breaking down through prior support.

Stocks

A share represents partial ownership of a company’s earnings and growth prospects. Zoomed out to an index, the stock market becomes a barometer for the economy’s expected earnings power. Corrections tend to show up technically as sharp expansions in volatility (visible in the VIX) followed by mean-reverting consolidation, while sustained bull trends tend to show a stair-step pattern of balance, breakout, new balance, breakout.

9. Following Institutional Positioning

Once a macro-technical thesis is formed, it helps to check whether large institutional players are actually positioned that way.

The weekly Commitment of Traders (COT) report, published by the CFTC, breaks down futures market positioning into commercial hedgers, large speculators (the “smart money”), and retail traders — offering a direct read on whether institutions are net long or net short a given market, and whether that positioning is expanding or unwinding week over week.

Upgrade to Paid

10. The Entry Models

Two entry models come directly out of the volume-profile framework described in Section 5:

1. The failed-auction re-entry (”break-in”). Price pushes below (or above) a recognized value area, fails to sustain the move, and closes back inside it — signaling that the prior flow of buyers or sellers is reasserting itself. This is typically the highest-probability, best-risk-to-reward entry, because the stop-loss can sit just beyond the failed excursion.

2. The breakout continuation (”break-out”). After a shift to a new value area, price retests the edge of that new range before continuing in the direction of the drift. This entry sacrifices some risk-to-reward for a higher confidence that the trend is already confirmed.

In ranging markets — common in forex outside of major policy shifts — the same volume-profile logic can instead support a mean-reversion approach, fading moves back toward the middle of an established range rather than chasing a breakout.

Upgrade to Paid

A practical checklist before pulling the trigger on either model:

• Does the higher timeframe trend agree with the direction of the trade?

• Is there a macro tailwind (rate expectations, real yields, risk sentiment) supporting the same direction?

• Does COT positioning support or at least not strongly contradict the trade?

• Is the entry occurring at a genuine high-volume node or value area edge, not an arbitrary round number?

• Is volume confirming the move (expanding on breakout, or drying up on the failed excursion)?

11. Risk Management and Position Sizing

No entry model survives without disciplined risk control. A few non-negotiable principles:

• Risk a fixed, small percentage of capital per trade — commonly 0.5% to 2% — regardless of how confident the setup feels.

• Let the stop-loss define position size, not the other way around. Identify the technical invalidation point first (just beyond the failed auction, or below the retested breakout level), then size the position so that a stop-out only costs the predetermined risk percentage.

• Use ATR to normalize stops across volatility regimes. A stop that’s appropriate in a calm range will be far too tight during a volatility expansion, and vice versa.

• Plan the reward before the trade, not after. A setup with an unclear or asymmetric reward relative to its risk isn’t a setup — it’s a guess.

• Correlated positions count as one trade for risk purposes. Being long two currency pairs both driven by dollar weakness is really one dollar bet, not two diversified ones.

  Upgrade to Paid

12. Building a Trading Plan

A repeatable process beats a brilliant one-off idea every time. A simple structure that ties everything above together:

1. Macro scan — check the current rate-hike/rate-cut expectations, inflation and employment trends, and overall risk sentiment.

2. Market selection — identify which asset class has the clearest macro tailwind right now (bonds, forex, gold, equities).

3. Structure mapping — mark the higher-timeframe trend, key value areas, POC, and VAH/VAL on the chosen instrument.

4. Positioning check — confirm COT data isn’t fighting the thesis.

5. Entry trigger — wait for either the failed-auction re-entry or the breakout retest, on the timeframe used for execution.

6. Risk definition — set the stop at the technical invalidation point, size the position from the fixed risk percentage.

7. Journal the trade — record the thesis, the entry logic, and the outcome, win or lose, to build a genuine statistical edge over time rather than a collection of anecdotes.

   Upgrade to Paid

13. Common Mistakes and How to Avoid Them

• Trading against the higher timeframe trend because a lower timeframe pattern “looks perfect.” The lower timeframe pattern is almost always noise relative to the dominant drift.

• Ignoring volume entirely and trusting a breakout on price alone. Price without volume confirmation is only half the story.

• Stacking correlated indicators and mistaking agreement between them for genuine confluence.

• Moving stop-losses further away mid-trade instead of accepting the original invalidation point was hit.

• Sizing positions based on conviction rather than the stop-loss distance, which turns a well-planned setup into an oversized bet.

• Fighting the macro backdrop — taking a technical long in an asset class with a clear structural macro headwind, and vice versa.

  Upgrade to Paid

14. Final Thoughts

Technical analysis, done well, isn’t about memorizing patterns — it’s about reconstructing where large, slow-moving capital is likely positioned and reacting, using volume and structure as the evidence trail, and using macro context to lean the odds in a consistent direction over many trades.

None of this guarantees a winning trade — no framework does — but it reframes trading around genuinely persistent, well-documented mechanics rather than short-lived retail patterns that tend to stop working as soon as enough people notice them.

Disclaimer: This article is for educational and informational purposes only and does not constitute financial advice. Trading involves substantial risk of loss, including the potential loss of principal. Always do your own research and consider consulting a licensed financial advisor before making investment decisions.

For two years I’ve been one of the people telling you Ethereum was the boring, sensible way to own the “internet of value.” Bitcoin was digital gold, fine, but Ethereum was the actual settlement layer. DeFi, stablecoins, tokenised treasuries, all of it runs through ETH. I liked that story. A lot of smart people still like that story.

I don’t anymore, and I want to walk you through exactly why, using the numbers rather than the vibes.

Subscribe now

The Chart Nobody Wants to Screenshot

Ethereum hit its all time high of roughly $4,953 in August 2025. As of this week it’s trading in the $1,700 to $1,800 range. That’s a drawdown of around 70 percent from peak, worse than Bitcoin’s roughly 51 percent slide from its own October 2025 high of $126,198.

Beta cuts both ways. Ethereum has always amplified whatever Bitcoin does, rallying harder in the good times and falling further in the bad ones. That’s not new information. What is new is how the fall is happening this time. It isn’t retail panic selling on Reddit. It’s institutional money, in broad daylight, in the ETF flow data, every single day.

The Number Everyone’s Ignoring

Read more

I want to tell you about a mistake I almost made in print.

When the first leaks about SpaceX’s Bitcoin holdings hit my feed, I had the post half-written in my head. Famous founder. Largest IPO in history. A pile of Bitcoin on the balance sheet. The headline practically wrote itself: “Smart money is quietly stacking sats.” I have written that exact kind of post before, and you have probably shared one. It feels good. It confirms what we already want to believe.

Then I sat down with the actual S-1 filing. And one line stopped me cold.

This piece is the version of the story I wish someone had handed me before I hit publish. It is built for the reader who is new to all of this and wants the plain English version, and for the desk that already trades crypto exposure and just wants the part everyone else skipped.

What SpaceX Actually Holds (The Plain Numbers)

Let us start with what is not in dispute, straight from the registration statement SpaceX filed ahead of its Nasdaq debut under the ticker SPCX.

SpaceX holds 18,712 Bitcoin. As of the filing date, that position carried a cost basis of roughly $661 million and a fair value in the neighborhood of $1.29 billion. Doing the simple math, the company paid an average of about $35,320 per coin, which means even after a brutal year for the asset it is sitting on a sizable paper gain. The filing classifies the holding as a “strategic reserve” and notes it represents only about 1.8% of total assets.

That last figure matters more than the headline number. This is not a company betting the farm. It is a company parking a sliver of its treasury in Bitcoin and getting on with the business of building rockets, satellites, and AI services.

For context, that 18,712 coins places SpaceX among the larger corporate holders, somewhere around eighth on the public-company leaderboard. It sits ahead of names like Coinbase and Riot, behind the runaway leader Strategy at roughly 845,000 coins, and just ahead of Elon Musk’s other company, Tesla, at 11,509. So yes, it is a real position. It is also, against the scale of Bitcoin’s roughly 19.8 million coins in circulation, a rounding error.

Hold that thought. We will come back to scale.

The One Line That Changed My View

Here is the part I almost left out of my original draft, because it did not fit the story I wanted to tell.

SpaceX did not simply accumulate and hold. The company began buying around 2021, and at one point its stack peaked near 28,000 coins. Then 2022 arrived, Bitcoin collapsed, and SpaceX cut its position by roughly 70%. The 18,712 coins it reported are what is left after that retreat.

Read that again, because it is the whole ballgame for anyone calling this “conviction.”

The bull narrative depends on a story about belief. The idea is that sophisticated operators look at Bitcoin, see something the rest of us are missing, and hold through the noise. But the filing tells a different and far more human story. SpaceX bought near the top of a cycle, watched the price fall apart, and sold a large chunk into the weakness, exactly the behavior you would expect from a treasurer managing risk rather than a true believer.

That is not a criticism of SpaceX. Trimming a volatile asset during a liquidity crunch is responsible corporate finance. But it is the opposite of the diamond-hands legend people are building around this disclosure. You cannot cite a holder as proof of unshakable institutional conviction when the public record shows they shook.

I learned this lesson the hard way years ago with a different “smart money is buying” story. I anchored on the buying and ignored the selling that came later, and my readers anchored with me. I am not doing that to you here.

So Is It Bullish? Here Is The Honest Answer

Let me give you the nuanced version, because a one-word answer would be doing you a disservice.

Mechanically, it changes almost nothing. Bitcoin’s market value sits north of a trillion dollars. SpaceX’s roughly $1.2 to $1.3 billion position is a fraction of one percent of that. No new coins are being bought as a result of the IPO. The holding has been static for a while, with no major purchases or sales reported around the listing. If your thesis is “this disclosure tightens supply and pushes price up,” the numbers simply do not support it.

There is even a short-term bearish case. Several analysts flagged the listing itself as a potential headwind. A historic IPO of this size pulls a lot of speculative capital toward one event. Bitcoin spent the days before the debut grinding sideways in the low sixty thousands, with sentiment readings sitting in “extreme fear” and the price down sharply from where it traded a year earlier. When the biggest show in town is a stock, money tends to leave the side tables. That is the opposite of a demand shock for the coin.

So where is the real, if modest, bull case? It is not about supply. It is about normalization. For years the only large public vehicles offering Bitcoin exposure were companies built specifically to hold it, the leveraged proxies whose share prices live and die on the gap between their market value and the value of their coins. SpaceX is a different animal. It is an operating company with real revenue from rockets and Starlink that happens to keep some Bitcoin next to its cash. When a business of that profile carries the asset through the largest listing in history and treats it as unremarkable treasury management, it quietly tells every CFO watching that this is now a normal, boring thing to do.

That signal is real. It is also slow, soft, and impossible to put a price target on. It is a “someday this matters” data point, not a “buy this week” catalyst.

The Trap The Treasury Company Bears Already Walked Into

If you spend time in the skeptical corners of Seeking Alpha and the research notes, you have seen the bear case against Bitcoin treasury companies, and it is a good one. The argument runs like this: these firms traded at large premiums to the value of their coins, funded more buying by issuing stock at those premiums, and the whole machine works in reverse the moment the premium fades. Reporting through late 2025 showed a meaningful share of the top hundred treasury names trading below the value of their holdings, with even the sector leader slipping to a discount at points. Critics called it a potential “spiral of doom,” where falling demand drags the share price down, which drags demand down further.

Here is the thing professionals need to hear clearly. Most of that bear case does not apply to SpaceX, and that is precisely why I refuse to wave it around as a reason to be negative here. SpaceX is not issuing stock to buy more coins. It is not selling a leveraged Bitcoin bet dressed up as an operating company. Borrowing the treasury-company critique and stapling it onto SpaceX would be lazy, and it would make the same error in reverse that the bulls are making: forcing a famous name into a story it does not belong to.

The discipline, for both sides, is the same. Look at what the entity actually is and what the filing actually says. Do not let the logo do your thinking.

What I Am Actually Doing With This

I promised you a trade, or at least a posture, so here it is in plain terms. None of this is personalized advice, and I am not your financial adviser, but this is how I am reading it on my own book.

I am not adding Bitcoin exposure because SpaceX owns some. That reasoning does not survive contact with the numbers. If I want Bitcoin, I will own Bitcoin directly or through a low-cost vehicle, not pay attention to a company whose coins are 1.8% of its assets and which already proved it will sell into a downturn.

I am watching the next few earnings cycles for one specific thing: whether SpaceX leaves the position alone, trims it, or adds to it. The filing is a snapshot. The behavior over the coming quarters is the actual signal. If a marquee operating company holds steady through volatility, that does more for the normalization story than the initial disclosure ever could. If it quietly trims again, the “conviction” crowd loses its last leg to stand on.

And on the macro asset itself, I hold the long-term frame loosely. There is a striking pattern in the history that the bears tend to skip, which is that Bitcoin’s average price across a full calendar year has, so far, never finished below the average from two years prior, even through its ugliest bear markets. That is not a promise. Patterns break. But it is the honest counterweight to the gloom, and leaving it out would make me guilty of the same selective storytelling I am warning you against.

The Takeaway

The fun headline was “SpaceX is stacking Bitcoin, follow the smart money.” The true headline is quieter and more useful. A major operating company holds a small, static Bitcoin reserve, already cut most of it once when things got rough, and just carried the remainder through a historic IPO without making a fuss. That is mild evidence that Bitcoin is becoming ordinary on corporate balance sheets. It is not evidence of a price catalyst, and it is not a license to chase.

I almost sold you the exciting version. I am glad I read the filing first.

Your turn. Did the “smart money is buying” narrative ever cost you on a trade, the way the SpaceX story almost cost me? Tell me in the comments, the messier the story the better. And if a friend forwarded you this, the subscribe button is right there. I read the filings so you can read the takeaways.

Nothing here is investment advice. Do your own work, size your own risk.

Let me tell you what I did last Tuesday.

I sat down with about six months of IMF reserve data, central bank purchase filings, trade settlement reports, and US Treasury flow data. I was supposed to be writing about something else entirely. Three hours later I had abandoned that piece completely because what I was looking at felt genuinely important, and I could not understand why it was not the lead story on every financial publication in the world.

Here is what I found. The people who actually run the global financial system, the ones who sit in the quiet rooms, who manage sovereign wealth funds and national reserve portfolios, who speak in the measured language of basis points and allocation frameworks, are quietly, systematically, and at an accelerating pace, moving away from the US dollar.

Not talking about it loudly. Not writing op-eds. Just buying gold. Relentlessly.

And the numbers behind that behavior tell a story that every investor, whether you manage $100 million or $10,000, needs to understand right now.

Subscribe now

The Number That Started All of This

The US dollar's share of global foreign exchange reserves has fallen below 57% for the first time since 1995, reaching 56.9% in late 2025 and continuing to fall from there. This is a structural decline from a peak of 72% back in 2001.

Read that again. From 72% to below 57% in roughly 25 years. A 15-percentage-point erosion in the world’s most important financial instrument.

Now here is the part that makes a macro trader sit up straight. That decline is not linear. It is accelerating. The drop from 60% to below 57% happened faster than any previous 3-percentage-point slide in the dataset. Whatever is driving this, it is not slowing down.

What Is Actually Happening Here

To understand this, you need to know the most important financial concept most people never learn in school: the exorbitant privilege.

Because the US dollar is the world’s reserve currency, the United States gets to do something no other country on earth can do. It borrows money in its own currency, at rates it partly controls, and the whole world essentially has to absorb that debt because they need dollars to buy oil, settle trade, and hold as national savings.

It is the financial equivalent of being the only casino in town. Everyone has to play by your rules. Everyone needs your chips.

That privilege is now, for the first time in living memory, visibly fraying.

The Debt That Broke the Camel's Back

In 2026, the federal government spends more on interest than on defense. That is a milestone reached in 2024 and now permanent under any plausible baseline. Within a decade, Brookings projects interest to consume more than 30% of federal revenues.

Let me contextualize that for you.

The US national debt has now crossed $39 trillion. Interest costs have surged because of a one-two punch: the federal debt load has ballooned by trillions while interest rates climbed sharply from near-zero pandemic lows. As a share of the economy, interest costs have doubled from 1.6% of GDP in 2021 to a record 3.2% in 2025.

In 2020, interest payments totaled $345 billion. Just six years later, they have nearly tripled, reaching over $1 trillion per year.

A trillion dollars a year. Just on interest. Not on roads, hospitals, defense, or anything that makes life better for anyone. Just the bill for borrowing money the country already spent.

The gross national debt crossed $39 trillion in March in less than five months after it hit $38 trillion, and the 2026 deficit is on track to hit 5.8% of GDP at a moment of relative peace and full employment.

That last part is the part that should scare you. These are not wartime deficits. These are not recession deficits. The US is running structural trillion-dollar-plus deficits in the middle of a growing economy at full employment. If a recession arrives, those numbers get horrifying very quickly.

Why Central Banks Are Doing What They Are Doing

Now you understand the why. Central banks around the world are watching this math and drawing their own conclusions. Their job is not to make a point or write an op-ed. Their job is to protect the purchasing power of their nation’s reserves across a 30-year horizon. And what they see is a creditor nation that has fundamentally changed its relationship with fiscal discipline.

So they are buying gold.

Gold hit an all-time high of $5,589.38 per ounce on January 28, 2026, a milestone that would have seemed unthinkable just a few years ago.

The current rally is distinguished by record central bank buying, with purchases since 2022 more than twice their 2015-19 average. Central banks' share of total gold demand rose to nearly 25% in 2024, compared with 12% in 2015-19.

An HSBC Reserve Manager Survey from 2026 documented that nearly 40% of reserve managers indicated plans to increase gold holdings during the year, with the majority explicitly noting that prevailing price levels were not a deterrent to their accumulation strategy.

Read that sentence carefully. Nearly 40% of reserve managers are buying gold in 2026 and the price does not matter to them. This is not speculative investment. This is sovereign-level portfolio restructuring.

Gold's share of global reserves has risen from 13% in 2017 to approximately 30% in 2025, with BRICS+ nations now holding 17.4% of global gold reserves, up from 11.2% in 2019.

The Trade Settlement Story Nobody Is Covering

Here is the piece of this that most financial media is genuinely missing.

BRICS nations now settle approximately 67% of intra-bloc trade in local currencies, up from under 20% a decade ago.

Think about what that number means. The BRICS bloc, which now represents a meaningful share of global GDP and population, has gone from doing nearly all of its trade in dollars to doing barely a third of it in dollars. In ten years. Quietly. Without a single emergency summit or breaking news chyron.

Some ships made payments in renminbi via China's CIPS payment system or in crypto-assets to transit through the Strait of Hormuz in early 2026.</cite> The Strait of Hormuz. The chokepoint through which roughly 20% of the world's oil flows. Transactions that, for half a century, happened exclusively in US dollars.

Reports indicate that over 90% of trade between Russia, India, and China now occurs without dollars, signaling a profound realignment in global trade flows.

This is not a protest. This is infrastructure being built in parallel. And infrastructure, once built, gets used.

I Want to Be Honest About What This Is Not

Here is where I want to slow down and be careful, because this topic attracts a lot of breathless commentary that does not hold up.

This is not the dollar dying tomorrow. It is not even dying in the next decade. The dollar remains by far the dominant invoicing currency for global trade. US Treasuries remain the deepest, most liquid financial instrument on earth. The Fed’s ability to act as the global lender of last resort is unmatched. When genuine panic hits markets, the dollar almost always strengthens, as we saw in 2020 and 2022.

What this is, more precisely, is a slow but measurable shift in the risk premium attached to holding dollars at the margin. It is sovereign reserve managers gradually diversifying away from a concentrated position they no longer fully trust, in the same way an institutional portfolio manager quietly trims a position that has gotten too large and too complicated.

The difference is that when 190 countries start doing that at the same time, the cumulative effect is enormous. And that is what the data says is happening right now.

What This Means for Actual Portfolios

Let me give you the practical part of this, because I know many of you are professionals who need to translate macro themes into positions.

Gold is not a trade anymore. It is a regime change. The buyers driving this rally are not momentum tourists or retail punters who will panic at the first 5% pullback. They are sovereign reserve managers with 30-year mandates who have told surveyors that price is not relevant to their buying decision. J.P. Morgan Global Research analysts expect gold to push toward $6,000 per ounce by year end. The structural bid does not go away because a number on a screen looks stretched.

The dollar’s weakness is not evenly distributed across assets. Commodities priced in dollars get a mechanical tailwind from a weaker dollar. Emerging market equities, particularly those with strong domestic revenue bases, benefit as the dollar loses purchasing power. Multinationals with large non-US revenue streams get a translation boost. This is not complicated, but it is frequently underappreciated in portfolio construction.

Treasuries face a longer-term structural headwind. The foreign official sector, which used to be a reliable buyer of US government debt, is shrinking as a share of the market. Foreign official holdings of US Treasuries have trended down from over 50% of the total market pre-2008 to around 30% by early 2025. That means the US has to increasingly finance its debt domestically, which creates upward pressure on yields and constrains the Fed’s room to maneuver.

The energy sector is a geopolitical pivot point. Any material expansion of oil trade settled outside the dollar, particularly from Gulf producers, would be the single most significant acceleration event for de-dollarization. Watch Saudi Arabia. Watch the UAE. Their reserve management choices are the canary.

The Moment I Actually Changed My Own View

I want to tell you something honest here. I have been broadly skeptical of de-dollarization narratives for most of my career. Every few years someone writes the “dollar is dying” piece and it ages poorly. I used to roll my eyes at it.

What changed my mind was not the geopolitics. It was not BRICS or the rhetoric around a new reserve currency. It was the US debt trajectory.

There is a difference between countries reducing dollar exposure because they are ideologically motivated to do so, and countries reducing dollar exposure because the arithmetic of US fiscal policy has genuinely changed their risk assessment. The first is political noise. The second is rational portfolio management.

What I see in the 2026 data is much more of the second. Reserve managers are not making speeches. They are making purchases. Quietly, consistently, and at record pace. That tells me something real has shifted in how the world’s most sophisticated institutional investors are thinking about the dollar over a 20 to 30-year horizon.

And if they are right, the asset allocation implications are significant enough that every serious investor should at minimum have a view on this, even if their view is that the dollar holds.

The Bottom Line

The dollar is not going to zero. The United States is not going to default. The global financial system is not going to collapse next year.

But the era of completely unchallenged dollar dominance, where the US could run any deficit it wanted and the world would absorb the paper, is ending in slow motion. The data is now unmistakable on this. Central banks are voting with their reserves. The IMF numbers confirm the trend. The trade settlement infrastructure is being built in parallel.

The investors who thrive in the next decade will not be the ones who panicked about the dollar. They will be the ones who quietly, systematically, and with clear logic, adjusted their asset allocation to reflect a world in which the dollar is still dominant but is no longer guaranteed.

That is a subtler call than “buy gold and hide.” But it is the right one.

This is not financial advice. I hold positions in assets discussed in this article. Always conduct your own research and consult a qualified financial advisor before making investment decisions.

What is your current dollar exposure, and has any of this changed how you think about it? Tell me in the comments. I read all of them.

TLDR: Harmonic has quietly transformed itself from a legacy video hardware vendor into the dominant software platform of the DOCSIS 4.0 upgrade cycle. With a $582M backlog, raised guidance, and a pending divestiture that completes the pivot, the market is only now beginning to reprice that shift.

The Setup: A Company the Market Is Still Mis-Filing

There’s a specific type of opportunity that tends to persist longer than it should: a business undergoing a genuine structural transition, where the old classification lingers in analyst models while the underlying fundamentals have already moved on.

Harmonic Inc. (NASDAQ: HLIT) sits squarely in that category right now.

For years, Wall Street filed Harmonic under “video technology,” a company making gear for broadcasters and pay-TV operators. Decent business. Modest growth. Not something that generates institutional urgency.

What those models missed (and many still miss) is that Harmonic has spent the better part of five years building the software infrastructure layer of the modern cable broadband network. The Q1 2026 results weren’t a sudden inflection. They were the visible consequence of compounding execution that started well before this year.

What Happened in Q1 2026: Reading the Numbers Carefully

Total Q1 2026 revenue came in at $171.8 million. That’s the consolidated top line, which includes the Video business, which is pending divestiture to MediaKind and should be treated as a discontinued operation for analytical purposes.

The number that matters is broadband segment revenue: $121.7 million, up 43% year-over-year, and well above the company’s own guidance range of $100–$105 million. That is a 16% beat at the midpoint, not a rounding error, but a signal that deployment activity is running ahead of management’s own visibility from just one quarter prior.

Non-GAAP EPS came in at $0.17, versus analyst consensus of $0.10, a 70% beat.

The guidance raise that accompanied the results is equally notable. Full-year 2026 broadband revenue guidance was lifted to $475–$495 million, from a prior range of $440–$480 million. Non-GAAP EPS guidance for the full year was set at $0.57–$0.67, above street consensus. Management does not raise guidance without improved deployment visibility; this is a statement about what’s already in the pipeline, not what they hope will materialize.

The metric that anchors the long thesis, however, is one that rarely appears in the headline summary.

Backlog and deferred revenue reached $582.1 million at quarter-end, up 87% year-over-year, with $307 million of that expected to convert within 12 months.

For context: the company is guiding to roughly $485 million in full-year broadband revenue. A backlog of $582 million does not mean 2026 is “already sold”; backlog can slip due to deployment delays, hardware availability constraints, and customer timing. But it does mean that revenue visibility is unusually high for a mid-cap infrastructure business, and that execution risk is shifting from “will the orders come?” to “can we convert what we’ve already won?” That’s a materially different risk profile.

Before We Go Further: What Harmonic Actually Does

If you’re new to the name, sixty seconds of context is worth the time, because the technology is the thesis.

Think about the cable company that delivers your internet. Historically, the “brain” of that network (the piece that takes broadband signals and routes them to millions of homes) was expensive, proprietary, purpose-built hardware. Large chassis. High maintenance costs. Slow to upgrade and difficult to scale.

Harmonic pioneered the alternative. Their cOS™ virtualized broadband platform moves that function into software, allowing cable operators to run their broadband infrastructure on commodity servers, deploy it flexibly across their networks, and scale capacity the way a software company scales: incrementally, without tearing out hardware.

The analogy to cloud computing replacing on-premise servers is imperfect but directionally useful. The architectural shift is real, and the industry is in the middle of making it.

DOCSIS 4.0 is the standard that is forcing cable operators to choose. The new specification offers up to 10 Gbps downstream and, critically, multi-gigabit symmetric upstream speeds. Every major operator in North America is now evaluating or deploying it. Harmonic is the platform most of them are choosing.

The Upgrade Cycle: Sizing the Opportunity Honestly

The DOCSIS 4.0 market is estimated at $2.24 billion in 2025, growing to $2.51 billion in 2026, and projected to reach $3.96 billion by 2030, representing a compound annual growth rate of approximately 12%, per industry research firm The Business Research Company.

For additional context: Dell’Oro Group forecasts that cable distributed access equipment revenue (the category Harmonic competes in most directly) will grow to $1.2 billion by 2028, as operators continue DOCSIS 4.0 and early fiber deployments. Cable access spending dropped sharply in 2024 and early 2025 as operators worked through excess inventory; the recovery from that trough is the setup.

The primary drivers of this cycle are fiber competition and subscriber capacity demands, not a single catalyst like AI. Fixed-line fiber alternatives from telecoms have been steadily gaining share in broadband subscriber counts. Cable operators that fail to offer multi-gigabit speeds will lose customers. That competitive pressure is structural and persistent, and it is the actual engine behind the upgrade cycle.

AI demand is a secondary accelerant worth acknowledging. Applications that require real-time, bidirectional data flows (voice agents, continuous syncing, cloud-rendered computing) do place greater demand on upstream bandwidth than prior-generation applications. As CEO Nimrod Ben-Natan noted on the Q1 2026 call, “AI is making broadband capacity and quality of service even more essential.” That framing is fair. But readers should weight fiber competition and capacity ceilings as the primary drivers, with AI demand as a directional tailwind rather than a core causal thesis.

The practical result for Harmonic: the capital spending freeze that suppressed results in 2024 and early 2025 is thawing, confirmed by Dell’Oro’s report that global broadband access equipment revenue grew 2% year-over-year in Q4 2025, marking the first year-over-year increase in several quarters.

The Moat: Why the 47% Share Position Is Durable

According to Dell’Oro Group, Harmonic holds 47% of the cable broadband infrastructure market, followed by CommScope at 34% and Vecima Networks at 10%.

Market share statistics are more useful when you understand why the share exists and what would have to change for it to erode. That is where the competitive analysis needs to go beyond the headline number.

Switching costs are high and multi-dimensional. A cable operator that deploys Harmonic’s cOS platform doesn’t just install software; they integrate it into their OSS/BSS stack, train their NOC teams on Harmonic’s management interface, build vendor relationships with Harmonic’s support organization, and align their roadmap to Harmonic’s feature releases. Migrating to a competitor’s platform mid-cycle requires re-integration, retraining, parallel-running two systems, and accepting deployment disruption during an upgrade period when uptime is critical. These are genuine friction points, not theoretical ones.

The software architecture creates compounding lock-in. As operators deploy more Harmonic nodes and accumulate more operational data inside Harmonic’s analytics layer, the platform becomes more valuable, not less. Harmonic has been building AI-powered network analytics into cOS, and those analytics are trained on each operator’s specific traffic patterns. Switching to CommScope means starting that data accumulation over.

CommScope’s structural challenges are real. CommScope is the primary competitor, and it is facing meaningful headwinds of its own: a debt-heavy balance sheet, a broader hardware portfolio requiring capital allocation trade-offs, and a product architecture that is less natively virtualized than Harmonic’s. CommScope is not going away, and 34% share is not trivial, but they are not in an offensive position. They are defending existing relationships while managing financial constraints that limit R&D investment.

Vecima’s 10% share is largely niche. Vecima competes effectively in specific mid-market deployments but lacks the scale and global support infrastructure to win Tier 1 operator contracts.

Technology proof points reinforce the lead. In July 2025, Harmonic set a new industry record at the CableLabs Interop event, with cOS delivering 14 Gbps downstream throughput across a multi-vendor network, demonstrating that their platform performs at the frontier of what the DOCSIS 4.0 standard can deliver. In September 2025, they completed the first Unified DOCSIS 4.0 production deployment on a live network with Mediacom. These are not marketing milestones; they are operational proofs that Harmonic’s architecture is executing where competitors are still certifying.

The Transformation: What “New Harmonic” Changes in the Model

In December 2025, Harmonic announced a binding offer from MediaKind to acquire its Video Business segment for approximately $145 million in cash. The transaction is pending close in Q2 2026.

This is worth pausing on, because it changes the analytical framework entirely.

Video was Harmonic’s legacy segment: slower growth, more hardware-dependent, and requiring management attention that competed with the broadband business. Selling it for $145 million cash is not just a balance sheet event; it is an organizational simplification that allows capital, engineering resources, and executive focus to concentrate entirely on the broadband platform.

The financial effect is meaningful. The $145 million in proceeds strengthens the balance sheet directly. The $2.3 million in quarterly stranded costs currently dragging operating profit disappears. And the re-rating that institutional investors apply to a pure-play software infrastructure business, versus a mixed hardware/software company, tends to be substantial.

Harmonic also authorized a $200 million share repurchase program and executed $43 million of that in Q1 2026 alone, repurchasing 4.2 million shares. For a ~$1.5 billion market cap company, that is an aggressive pace of capital return, a signal from management that they regard the current share price as undervalued relative to the trajectory they see internally.

Customer Diversification: The Structural De-Risking That’s Underappreciated

For years, the Harmonic bear case could be summarized in one sentence: “It’s a two-customer story.” Comcast and Charter have historically driven the majority of broadband revenue, meaning any capex pause from either operator flows almost immediately into Harmonic’s results.

That risk has not disappeared. In Q1 2026, the top two customers still accounted for 58% of broadband revenue ($71.1 million of the $121.7 million total).

But Rest-of-Market revenue reached $50.6 million in Q1 2026, up 78% year-over-year, and now represents 42% of total broadband revenue. The diversification is happening, and the pace matters.

What is driving it? Three forces: DOCSIS 4.0 adoption among mid-tier and regional operators who previously lagged the Tier 1s; Harmonic’s fiber solutions (exemplified by the DNA Finland SeaStar optical node win in May 2026) opening doors outside the cable-specific market; and the cOS platform’s architecture becoming increasingly accessible to smaller operators who couldn’t previously justify the integration complexity.

The thesis, properly framed, is not that concentration risk has gone away. It’s that the trajectory is clearly reducing it, and that the 78% Rest-of-Market growth rate in a quarter where total broadband growth was 43% tells you the diversification is running ahead of the core business.

The Bear Case: A Fair Hearing

Credible investment analysis requires treating the opposition seriously. Here is the bear case stated as its proponents would state it.

Customer concentration remains the primary risk. 58% revenue from two customers is a binary exposure. Comcast’s DOCSIS 4.0 deployment timeline is subject to regulatory, competitive, and financial pressures that Harmonic cannot control. A single quarter of Comcast capex softness produces an immediate and visible revenue impact. This is not theoretical.

Gross margins are under pressure. FY2026 gross margin guidance of 50–51.5% is below historical levels, compressed by memory costs and new product ramp expenses. For a business pitching a software-centric premium multiple, margin trajectory matters. The current guidance implies gross margin headwinds persist through the upgrade cycle’s early stages.

Hardware supply chain risk is real. There are disclosed concerns about server availability that could delay deployments and affect revenue recognition timing. The revenue recognition model (which recognizes revenue upon deployment) means a hardware bottleneck upstream creates lumpiness even against a backdrop of strong demand.

The video sale hasn’t closed. The MediaKind transaction is pending French employee works council consultation and regulatory approvals. Until it closes, $2.3 million in quarterly stranded costs remain on the income statement, and there is headline risk if the transaction is delayed into H2.

Wolfe Research maintains an Underperform. Their thesis centers on near-term execution risk and the argument that the market is pulling forward a re-rating that has not yet been fully earned by multiple quarters of consistent execution. Wolfe standing alone as a bear is a signal worth taking seriously rather than dismissing.

What Actually Breaks This Thesis

Beyond the standard risk disclosure, it is worth being explicit about the specific scenarios that would materially change the investment case. These are not tail risks; they are plausible outcomes that warrant monitoring.

Scenario 1: Comcast pauses DOCSIS 4.0 capex. Comcast is Harmonic’s largest customer. If Comcast publicly revises its infrastructure capex guidance downward (due to subscriber growth pressure, rising financing costs, or a strategic pivot toward fiber-over-coax rather than DOCSIS 4.0), the revenue impact is direct and immediate. This is the single-largest binary risk in the thesis.

Scenario 2: CommScope closes the technology gap. The bear case on competitors has a time dimension. If CommScope resolves its balance sheet constraints, either through asset sales or refinancing, and redeploys capital into accelerated vCMTS development, the switching cost moat described above is less permanent than a static market share chart implies. Watch CommScope’s R&D spend and new customer win announcements as early indicators.

Scenario 3: Gross margins fail to recover. If memory costs and new product ramp expenses persist beyond the current year, the 50–51.5% gross margin range becomes the new floor rather than a transitional trough. A sustained margin compression would undermine the premium multiple argument and push intrinsic value estimates materially lower.

Scenario 4: The MediaKind transaction is delayed or renegotiated. The deal involves French regulatory process, which introduces timing uncertainty. A delay into H2 2026 prolongs stranded cost drag and defers the pure-play narrative crystallization. A price renegotiation, even a modest reduction from $145 million, would be a negative signal about Video asset quality.

These are the four scenarios to monitor. They are not predictions, but they are the specific events that warrant re-evaluating the position.

Valuation: Framing the Range Honestly

Valuation in a cyclical infrastructure company transitioning from two-segment to pure-play is genuinely difficult to precision-model. We offer several frameworks, each with different assumptions, and let readers weigh them.

Earnings-power framework. Management guided full-year 2026 non-GAAP EPS to $0.57–$0.67. A peer-group multiple of 25–30x forward earnings, appropriate for a software-centric infrastructure business with demonstrated market leadership and a long upgrade cycle ahead, produces an implied price range of roughly $14–$20. At the current price in the mid-$17s, this framework suggests the stock is approximately fairly valued to modestly undervalued under base-case assumptions. The upside comes from two sources: EPS above the guided range (possible given Q1’s pace), and a multiple re-rating as the pure-play narrative becomes consensus.

DCF framework. A DCF analysis using a WACC of approximately 9.5% (reflecting mid-cap tech risk, modest leverage post-divestiture, and the current rate environment) and a terminal growth rate of 3.5% (conservative relative to the 12.1% sector CAGR, but appropriate for a business with cyclical capex dependence) produces an intrinsic value estimate in the range of $22–$28 per share. The midpoint of that range (call it $25) implies roughly 45–50% upside from current levels. DCF models for cyclical businesses carry wide confidence intervals; we present this as a directional anchor, not a precision target.

The reconciliation. The earnings-power framework ($14–$20) and the DCF ($22–$28) are not contradictory. The gap reflects the horizon. The P/E analysis is priced on 2026 guided earnings, which are first-year pure-play numbers with stranded cost drag and margin pressure. The DCF captures normalized earnings power in years 3–10, after the upgrade cycle matures, margins recover, and the pure-play operating model reaches full efficiency. The $22–$28 range is not a 12-month target; it is what the business is arguably worth on a through-cycle basis. The nearer-term price target for the next 12–18 months is $20–$24, which requires both EPS execution at the top of the guided range and a moderate multiple expansion from the current re-rating in progress.

The Cross-Asset Angle: An Infrastructure Play in a Crowded AI Trade

The direct AI infrastructure trade (NVIDIA, the hyperscalers, custom silicon) is among the most crowded in global equity markets. The valuations in those names require sustained, near-perfect execution over multiple years to justify.

The broadband infrastructure layer is a different exposure. It is not a bet on which AI application wins or which model architecture dominates. It is a bet on the physical network that all AI-integrated applications (voice agents, cloud rendering, continuous sync) depend on to reach end users.

Cable operators are upgrading whether AI demand materializes at the pace bulls expect or not. They are upgrading because fiber competitors are taking subscriber share, because their own subscribers are demanding higher speeds, and because their networks are approaching throughput ceilings on legacy DOCSIS 3.1 architecture. These are structural drivers with their own momentum.

The picks-and-shovels framing is appropriate here. Harmonic doesn’t need to pick the AI application winner. It just needs cable operators to keep deploying, and the competitive and capacity pressures that are driving that deployment are not dependent on a single variable.

The Trade

Positioning: We view HLIT as a medium-conviction position appropriate for investors with a 12–18 month horizon and tolerance for mid-cap volatility. Entry in the $13–$17 range remains reasonable given the thesis intact. The 12–18 month price target is $20–$24, with a through-cycle DCF range of $22–$28 for investors willing to hold across multiple quarters of execution.

Catalysts:

• MediaKind close (Q2 2026): Eliminates stranded costs, delivers $145M cash, completes the pure-play narrative for institutional buyers.

• Q2 2026 earnings (August 2026): The critical validation quarter. If Rest-of-Market growth sustains above 50% and gross margins show sequential improvement, customer concentration risk structurally abates and re-rating logic strengthens.

• Backlog conversion pace: Watch $307M expected to convert within 12 months as the leading indicator. Deferred revenue drawdown in each quarterly filing is the most direct tracking signal for deployment momentum.

• Comcast / Charter milestones: Public announcements of accelerated DOCSIS 4.0 rollout timelines from either of the top two customers are direct readthroughs to Harmonic revenue timing.

• International wins: Rest-of-World broadband revenue grew 33% in FY2025 and is now 41% of total broadband. Each new international operator win (Vyve, DNA Finland) extends the TAM narrative and reduces geographic concentration.

Position sizing: Mid-cap, single-name infrastructure play with real customer concentration risk. A 2–5% allocation is consistent with most portfolio construction frameworks at this conviction level. The asymmetry is favorable, but concentration tail risk is real and should be sized accordingly.

Thesis invalidation signals: A Comcast capex reduction statement; a renegotiation or delay of the MediaKind transaction beyond Q3 2026; gross margins printing below 49% in Q2; CommScope announcing a major Tier 1 operator win on a new platform architecture.

The Bottom Line

Harmonic has spent five years building a platform position in a market that is now in early-stage deployment ramp. The 47% market share, the $582 million backlog, the 78% Rest-of-Market growth, and the pending transformation into a pure-play broadband business are all pointing in the same direction.

This is not a story without risk; the customer concentration, margin dynamics, and supply chain constraints are real and deserve ongoing attention. But the evidence from Q1 2026 is that execution is running ahead of guidance, diversification is accelerating, and the re-rating from “legacy video company” to “pure-play broadband infrastructure platform” is underway.

Markets re-rate slowly. The analytical edge here is in seeing the new framework before the consensus fully prices it in.

There is a question circulating in every serious macro conversation right now: the AI spending cycle is real, the numbers are enormous, and yet a significant portion of the equity market’s exposure is concentrated in a handful of chip and hyperscaler names that everyone already owns.

The rotation thesis gaining traction in institutional circles is this: the next leg of AI investment flows not into software or models, but into the physical substrate that makes AI possible at scale - the power, the thermal management, the networking, and eventually the embodied systems that take intelligence off the screen and into the physical world.

This piece maps that thesis precisely: what the data actually shows, where the valuation math works, where it breaks, and what the trade looks like across different risk profiles.

Part I - The Scale of the Commitment

The starting point is the raw capital commitment, because it sets the context for everything else.

The five largest US cloud and AI infrastructure providers - Microsoft, Alphabet, Amazon, Meta, and Oracle - have collectively disclosed capex plans of $660–$690 billion for 2026. Amazon alone is projecting $200 billion, up from $131 billion in 2025. Alphabet is guiding to $175–$185 billion, roughly double its 2025 figure. Meta is at $115–$135 billion. Microsoft is tracking toward $120 billion or more.

Goldman Sachs Research has noted that analyst estimates have consistently underestimated hyperscaler capex in recent cycles. The Wall Street consensus estimate for this group’s 2026 spending began the year at $465 billion and has since been revised to $527 billion - an upward revision of more than $60 billion in a single earnings season. Every quarter, the number moves higher.

The capital flowing into the physical layer of this infrastructure - not the models, not the software - breaks into three categories: GPU and custom chip procurement, physical data center construction, and networking and cooling infrastructure. It is the second and third categories where the rotation opportunity concentrates.

The longer-term figures are harder to dismiss. McKinsey estimates that companies will invest close to $7 trillion in global data center infrastructure by 2030. To put that in context: Germany’s 2025 nominal GDP is approximately $5 trillion, and Japan’s is approximately $4.3 trillion - their combined output is around $9.3 trillion. The $7 trillion infrastructure estimate represents capital expenditure equivalent to roughly 75% of those two economies’ annual output. That is a materially different comparison from versions of this claim that have circulated loosely in research notes; the correct framing matters.

Part II - The Real Constraint, and Why It Changes the Stock Map

For the past two years, the primary constraint on AI deployment was GPU supply. Nvidia’s H100 and then Blackwell architectures were the scarce resource. Whoever could secure allocation was building; everyone else was queuing.

That constraint is easing. Nvidia now reports visibility into at least $1 trillion in AI systems demand through the 2026–2027 period, and supply capacity is being built to match. The chip bottleneck is not eliminated, but it is no longer the binding constraint.

The binding constraint has shifted to power delivery.

As of early 2025, roughly 190 gigawatts of hyperscale data center capacity had been announced globally across hundreds of disclosed projects. The structural problem is not construction timelines - a data center can be built in 12 to 18 months. The problem is grid interconnection. Connecting a new large-scale facility to the US electricity grid through standard utility processes currently takes five to seven years on average. The construction cycle and the power delivery cycle are misaligned by a factor of three to four.

This creates a specific investment opportunity: the companies that control power now, or that have found structural ways to bypass the interconnection queue, are sitting on a scarcity premium that is not yet fully priced into most utility-sector frameworks.

US data center power demand is projected to reach roughly 750 terawatt-hours annually by 2030, representing a near-tripling of current capacity. Data centers, which currently consume approximately 6% of total US electricity, are projected to reach close to 11% by 2030. That shift will reorder utility planning and energy policy for the better part of a decade.

Part III - The Stock Map, Layer by Layer

Layer 1: Power Generation

The cleanest listed expression of the AI power thesis in US equities is Constellation Energy (CEG). Constellation operates the largest nuclear fleet in the United States, with approximately 90% of its output carbon-free. It has signed 20-year power purchase agreements with Microsoft (the Crane Clean Energy Center restart, supported by PJM grid capacity) and Meta (the Clinton Clean Energy Center, 1,121 megawatts). The company has submitted 5,000 megawatts of new capacity to PJM and is developing what it terms “powered land” - shovel-ready data center sites co-located directly with nuclear plants, structured to bypass the congested public interconnection queue entirely. The company guided capex of $410–$580 million for 2026, with management flagging EPS upside of $0.40–$1.00 per gigawatt of additional nuclear PPA signed. The stock trades around 26x forward earnings - elevated relative to traditional utility comps, but below most growth industrials with comparable revenue visibility.

Vistra Corp (VST) offers a different structure. Its acquisition of Energy Harbor added a substantial nuclear fleet, which it operates alongside a large natural gas portfolio - the fuel type hyperscalers are actually contracting today while nuclear restart timelines extend. VST trades at roughly 18x forward earnings, meaningfully cheaper than CEG, but carries approximately $19.6 billion in net debt. For investors who want more operational leverage to the AI power cycle and are comfortable with the balance sheet, Vistra provides a higher-beta entry to the same structural trend.

The specific risks that belong in any honest write-up: PJM is experiencing significant interconnection queue delays, and some disclosed data center tie-ins have had projected grid delivery dates pushed to 2030–2031. CEG’s near-term PPA pipeline is concentrated among a handful of hyperscaler counterparties; any meaningful pullback in Big Tech capex directly impairs the contracting backlog. Nuclear license extension processes introduce regulatory timing risk that cannot be hedged.

Layer 2: Thermal Management and Electrical Infrastructure

The shift from air-cooled to liquid-cooled data center architectures is not a future event - it is happening in current construction. Blackwell-class GPU racks operate at power densities that air cooling cannot adequately manage; direct-to-chip or immersion liquid cooling is required to run at nameplate capacity. Industry estimates put liquid cooling adoption at under 5% of new builds in 2023, rising to a projected 35–40% of new builds in 2026. Every data center designed for the 2026–2028 delivery window is being engineered for liquid cooling from the foundation up.

Vertiv Holdings (VRT) is the most direct listed proxy for this transition. The company sells power distribution, uninterruptible power supply systems, and thermal management infrastructure - the equipment layer between grid connection and operational GPU rack. Full-year 2024 results showed organic sales growth of 26% year-over-year, adjusted diluted EPS growth of 47%, and Q4 2024 organic orders up approximately 252% compared to the prior year quarter - an orders acceleration that is the leading indicator the market is watching. The reported project backlog has exceeded $15 billion, roughly 12 to 18 months of forward revenue at current run rates. Management’s 2026 guidance calls for net sales of $13.25–$13.75 billion with organic growth of 27–29%, and adjusted diluted EPS of $5.97–$6.07 - representing approximately 43% EPS growth at the midpoint compared to full-year 2024.

A direct response to the valuation question: At current levels, Vertiv trades at approximately 50x trailing earnings and roughly 32x forward 2026 EPS guidance. That is an elevated multiple for an industrial hardware business. Here is the math that either justifies or condemns it.

Consensus estimates project EPS of $6.16 in 2026, $8.13 in 2027, representing approximately 32% compound annual growth from 2026 forward. If you assume the AI capex cycle sustains through 2028 and VRT delivers on those estimates, you are paying roughly 19–21x earnings on a 2027 basis. If VRT re-rates to a 25x forward multiple - still a premium to industrial peers, but below today’s level - that implies a stock price in the $200 range by mid-2027. That is roughly 25% upside over 18 months from current levels, before any estimate revisions.

The risk is multiple compression, not earnings. If the AI capex cycle shows any sign of decelerating - through hyperscaler guidance cuts, ROI disappointment, or macro deterioration - VRT could re-rate from 50x to 25x trailing earnings while earnings are still growing, and the stock falls 40–50% despite a fundamentally healthy underlying business. That is the specific scenario you must underwrite before building a position.

Eaton Corporation (ETN) addresses the electrical infrastructure layer - transformers, switchgear, and power distribution between the grid and the rack. Approximately 70% of revenue ties to electrical segments. Electrical Americas rolling 12-month orders rose 42% year-over-year as of the most recent reporting, with total electrical backlog up 48%. Eaton has announced plans to expand into liquid cooling capability through targeted acquisitions, extending its portfolio further into the data center thermal chain. The company offers meaningfully lower valuation risk than VRT - trading near $377 against a Bernstein price target of $428 - and serves as the more conservative expression of the same data center buildout thesis.

Arista Networks (ANET) owns the networking layer. When thousands of GPUs communicate in a training cluster, they require ultra-low-latency, high-bandwidth Ethernet switching - Arista’s core product. Networking accounts for roughly 10–15% of total AI cluster cost, but it is operationally mission-critical; customers pay a quality premium rather than compromise on latency. Meta and Microsoft are anchor customers. Trailing revenue growth has run at 20–25% annually, and consensus projects continued growth in the 25–30% range through 2026 as AI cluster networking spend accelerates. The valuation is high; the risk, identically to VRT, is multiple compression if the capex cycle slows before estimates reflect it.

Layer 3: Embodied Intelligence - The Earliest and Highest-Risk Layer

The preceding two layers are infrastructure that exists today, is generating revenue today, and has visible demand backlogs extending 18–24 months. This layer is different: it is earlier-stage, less financially validated, but represents the largest potential expansion of the physical AI addressable market.

Physical AI - the deployment of machine intelligence into autonomous systems operating in the real world - represents a structural extension of the AI thesis beyond data center infrastructure. When AI operates in a warehouse, a car assembly line, or a logistics network, the addressable market is no longer knowledge work and software productivity. It is the physical economy.

The commercial deployment picture as of mid-2025 is real but nascent:

Figure AI, backed by OpenAI, Nvidia, Amazon, and Microsoft, is running its Figure 03 humanoid at BMW’s Spartanburg plant, having handled over 90,000 parts in a documented production pilot. The company holds a private valuation of approximately $39 billion as of late 2024.

Boston Dynamics’ electric Atlas has all 2025 production units committed to Hyundai’s Robotics Metaplant Application Center and Google DeepMind, with commercial availability for external customers targeted from 2027. Boston Dynamics’ implied valuation has moved dramatically - from $1.1 billion at Hyundai’s 2021 acquisition to IPO projections suggesting an $85 billion or higher target by Korean securities analysts, driven by the shift from hydraulic to fully electric Atlas architecture.

Tesla’s Optimus program is at a pre-commercial stage, with Elon Musk projecting 100,000–300,000 unit production at $20,000–$30,000 per unit at full scale - which, if achieved, would be the first genuinely affordable general-purpose humanoid. That production scale is a 2027–2028 story at the earliest. The valuation implication for Tesla as a company depends entirely on whether you believe the manufacturing cost targets and the speed of the ramp.

Chinese manufacturers are a structural consideration that most Western-focused analysis underweights. Unitree shipped over 5,500 humanoid units in 2025 and is targeting 10,000–20,000 in 2026. Its G1 sells at $16,000 today - already below Tesla’s long-term target price. Chinese manufacturers hold an estimated 90% of global humanoid unit volume. The same cost-curve dynamics that reshaped solar panel and EV component economics apply here, and Western investors building valuation frameworks around premium hardware pricing should model a Chinese competitive response explicitly.

The meaningful commercial threshold - multiple manufacturers deploying at scale into industrial customers - is a 2027–2028 event, not a 2025–2026 event. The capital is being deployed now. The valuations are being set now. The question is whether you want to pay 2027 prices in 2025.

Part IV - The Bear Case, Stated Without Softening

Any thesis with this level of consensus support demands serious stress-testing. Here are the three most credible objections, stated as forcefully as their proponents would state them.

The capital misallocation scenario. The infrastructure is not overbuilt - it cannot be, because grid interconnection queues prevent delivery fast enough for that outcome. The real risk is something more insidious: stranded assets. Hyperscalers commit to a data center in a location, begin construction, and then discover the grid connection will not be available until 2030. The building is real, the depreciation begins, but the revenue-generating compute cannot be installed. This is not hypothetical - PJM has flagged projected in-service delays for multiple disclosed projects. Capital deployed into data center facilities that cannot be powered on schedule earns zero return on that capital until the grid constraint resolves. The question for investors is whether the hyperscalers’ balance sheets are strong enough to absorb that timing mismatch - which, for Microsoft, Google, and Amazon, they almost certainly are, but for debt-funded second-tier players like Oracle, the math is more fragile.

The Cisco scenario at the infrastructure layer. Cisco sold the most essential networking equipment of the 1990s internet build-out and lost 86% of its market value between 2000 and 2002 - not because the internet failed, but because it overbought at peak cycle on over-extrapolated demand. The relevant companies here are not the hyperscalers; their balance sheets are self-funding. The risk is to the equipment suppliers with elevated multiples: VRT, ANET, ETN. If hyperscaler capex guidance cuts even one or two quarters - not cancellation, just a pause - multiple compression on 50x earnings stocks can be severe and fast. The Seeking Alpha bear case on MSFT makes this point explicitly: record capex and rapid GPU architecture turnover (cycles of 18–24 months) create the possibility of meaningful asset obsolescence before ROI materializes.

The ROI gap and its second-order effects. Pure-play AI vendors - OpenAI, Anthropic, and others - are posting rapid revenue growth, but their combined revenues remain a fraction of the infrastructure being deployed on their behalf. The hyperscalers are, in a meaningful sense, betting on a customer base that has not yet proven it can pay for the infrastructure being built for it. Goldman Sachs Research has observed that investors are already rotating away from AI infrastructure companies where capex is debt-funded and operating earnings growth is under pressure, rewarding instead those demonstrating a clear link between capex and revenues. If AI revenue growth at the application layer disappoints versus the infrastructure build-out rate, the rationale for $700 billion in annual capex comes under genuine pressure.

Part V - The full setup - entry level, stop-loss, and the smart money flow data behind this

Upgrade Now 50% off for 48 hours only

Read more

Shopify is no longer a bet on online storefronts. It is becoming a tollbooth on digital commerce itself.

That distinction matters because the market still values Shopify like a premium SaaS platform rather than what it increasingly resembles: a financial infrastructure layer with embedded AI monetisation and expanding enterprise gravity. The paradox is that investors continue debating whether Shopify deserves a “high multiple,” while the company is quietly evolving into something structurally more durable than software.

The consensus narrative says Shopify is expensive.

The contrarian argument is that the market still misunderstands what Shopify is becoming.

The Shift From Software Vendor to Commerce Infrastructure

For most of the last decade, Shopify was viewed through a narrow SaaS lens: subscription revenue, merchant growth, app ecosystem expansion. That framework now looks incomplete.

The company’s real transition is from “website builder” to operating system for commerce.

That sounds like marketing language until you examine the revenue architecture.

Subscription revenue still matters because it creates recurring cash flow and embeds merchants into the ecosystem. But the economic engine increasingly sits inside Merchant Solutions, including payments, capital, advertising, fulfilment integrations, and AI enabled optimisation tools.

That distinction changes the entire valuation debate.

Traditional SaaS businesses are constrained by seat expansion or pricing increases. Shopify scales differently. As merchant GMV rises, Shopify monetises the flow of commerce itself.

The more successful merchants become, the more Shopify earns.

That is a structurally superior model.

Payment penetration climbing from roughly 60% to 64% in twelve months illustrates the point. Every incremental percentage point shifts more transaction volume into Shopify’s financial rails. At scale, this creates operating leverage that most software companies cannot replicate.

Investors still anchored to legacy SaaS comparisons are likely underestimating how powerful this becomes over a 10 to 20 year horizon.

AI Is Not a Feature Layer

It Is a Margin Expansion Engine

Most public companies are currently “adding AI.”

Shopify appears to be rebuilding itself around it.

That difference is important.

The market has become numb to AI announcements because many are cosmetic: copilots nobody uses, chatbots with weak monetisation, or vague productivity promises disconnected from revenue.

Shopify’s approach is more economically tangible.

Tools like Sidekick AI and Shopify Magic directly influence merchant outcomes: faster product launches, automated customer support, improved fraud detection, conversion optimisation, and merchandising efficiency.

This matters because Shopify monetises merchant success.

If AI increases merchant sales conversion, Shopify captures a percentage of that uplift through Merchant Solutions revenue. If AI reduces operational friction, merchants stay longer and spend more across the ecosystem.

The monetisation loop is unusually clean.

And unlike most AI narratives, Shopify possesses something genuinely scarce: proprietary commerce data.

The company’s ecosystem spans millions of stores, millions of products, and hundreds of billions in annual GMV. That creates a commerce specific training environment that general purpose models cannot easily replicate.

This is where the moat begins widening.

Competitors like Wix or BigCommerce can integrate generic AI tooling. But Shopify’s advantage increasingly comes from behavioural commerce intelligence accumulated across the ecosystem itself.

That data flywheel compounds.

The longer merchants operate inside Shopify’s infrastructure, the more predictive the platform becomes.

And prediction is monetisable.

The overlooked detail is margin structure. AI credits, premium automation layers, and merchant productivity tooling are likely to carry exceptionally high gross margins relative to legacy commerce infrastructure. If Sidekick evolves into a meaningful paid layer across even a fraction of Shopify’s merchant base, investors may discover the company has quietly added a second software business on top of its transaction business.

That possibility is not fully reflected in current expectations.

The Enterprise Migration Story Is Still Early

One of the market’s persistent mistakes is assuming Shopify remains primarily an SMB platform.

That thesis is becoming outdated.

The migration upmarket is accelerating.

Large brands increasingly care less about owning fragmented custom infrastructure and more about speed, interoperability, checkout optimisation, and omnichannel integration. Shopify has become increasingly credible in those areas.

Enterprise partnerships matter not merely because they add revenue, but because they reshape perception.

When global brands choose Shopify, they validate the platform’s durability.

That credibility compounds.

Once major consumer brands standardise on Shopify infrastructure, procurement resistance weakens across the broader enterprise landscape. The platform stops being viewed as a startup tool and starts being viewed as default infrastructure.

This is precisely how platform dominance deepens.

The market likely still underestimates the long tail implications of enterprise adoption because investors remain overly focused on quarterly merchant additions rather than infrastructure standardisation.

Infrastructure businesses win slowly, then all at once.

Why The Competitive Moat Is Harder To Disrupt Than Bears Assume

There is a recurring bearish argument around Shopify:

“What stops another platform from copying this?”

Theoretically, very little.

Practically, almost everything.

The moat is not a single feature. It is the interaction between ecosystem density, merchant trust, payments infrastructure, developer incentives, switching costs, and proprietary commerce data.

That combination is difficult to recreate simultaneously.

A rival platform would need to:

• Match Shopify’s payments infrastructure

• Rebuild a 16,000+ app ecosystem

• Convince merchants to migrate

• Replicate years of behavioural commerce data

• Establish enterprise credibility

• Offer superior checkout conversion economics

All at once.

That is not impossible. But it is extraordinarily expensive and time consuming.

Meanwhile Shopify continues compounding scale advantages.

The developer ecosystem reinforces merchant adoption. Merchant adoption reinforces app creation. More transaction volume improves payments economics. Better economics improve merchant retention.

The flywheel feeds itself.

Importantly, Shopify also benefits from a form of psychological lock in. For many entrepreneurs and scaling brands, Shopify is now the default starting point for commerce.

Defaults matter more than investors often realise.

Once a platform becomes institutionalised in startup culture, agencies, venture ecosystems, and enterprise commerce teams, displacement becomes far harder than feature comparisons suggest.

The Valuation Debate Misses The Duration Of The Runway

This is where the contrarian case becomes strongest.

Yes, Shopify still trades at premium multiples.

But premium multiples attached to structurally compounding infrastructure businesses are often misunderstood because investors anchor too heavily to near term earnings compression.

The key question is not whether Shopify looks optically expensive on trailing earnings.

The real question is whether the market is accurately pricing the duration of its growth runway.

That runway increasingly spans:

• AI enabled commerce tooling

• Payments penetration expansion

• Enterprise migration

• Advertising monetisation

• Conversational commerce

• Embedded financial services

• Global merchant expansion

• Offline and omnichannel commerce integration

This is not a business approaching saturation.

It is arguably still in the early stages of ecosystem monetisation.

The market’s skepticism partly stems from macro concerns around consumer spending and SMB fragility. Those risks are real. Shopify is not immune to economic cycles.

But cyclical volatility is different from structural deterioration.

And structurally, Shopify appears stronger than it did three years ago.

The company generates meaningful free cash flow, maintains significant strategic flexibility, and continues expanding its role inside the commerce stack. That combination limits downside relative to earlier stages of its lifecycle.

Ironically, much of the “valuation risk” may already have been absorbed through multiple compression over the last several years.

The business improved while sentiment cooled.

That divergence is often where long duration opportunities emerge.

The Real Bull Case

The most important point is not whether Shopify grows 25% next year.

It is whether Shopify becomes indispensable infrastructure over the next two decades.

If it does, current valuation debates may eventually look small in hindsight.

The strongest compounders rarely appear cheap in real time because markets struggle to price optionality before it becomes visible in financial statements.

Amazon looked expensive for years.

So did Adobe.

So did Nvidia.

The common thread was not short term valuation compression. It was infrastructure dominance combined with expanding monetisation layers.

Shopify increasingly fits that pattern.

The company is no longer simply helping merchants build websites.

It is positioning itself as the coordination layer between AI, payments, logistics, advertising, and digital commerce itself.

If that thesis proves correct, today’s valuation may ultimately reflect the beginning of Shopify’s infrastructure era rather than the end of its growth story.

Final Asymmetry Assessment

Hypothesis: Shopify at $95-110 is a favorable asymmetric bet for investors with a 12-24 month horizon.

Supporting evidence:

• Low valuation floor: $75-85 (moderate bear), $50-60 (severe bear). Multiple de-rating already occurred (trading 50% below 6-year average).

• High ceiling: $140-180 (bull case). Catalysts are real, achievable, and partially priced in.

• Favorable risk/reward: 40-80% upside vs. 15-40% downside over 12-18 months.

• Margin of safety: Current P/E of 95-119x is within 1-2 std dev of fair value assuming 25-30% growth persists.

  

Critical condition for thesis: Growth must remain 20%+ through 2026. If growth decelerates to <15%, the bear case activates and upside evaporates. Monitoring Q2 2026 earnings and guidance will be key.

Subscribe now

Most investors think they understand Palantir.

They don’t.

They hear “AI”, “data analytics”, maybe “government contracts”… and they put it in a mental box next to every other tech company.

That mistake is exactly why most people will misprice this business.

Because Palantir isn’t playing the same game.

It’s not trying to sell software in the way you’re used to.

It’s trying to become something much deeper:

The operating layer that sits underneath how governments and corporations actually make decisions.

And once you see it that way… everything changes.

Read more

Upgrade to paid

Something doesn’t add up.

A company printing $81 billion in quarterly revenue… growing double digits… expanding margins… sitting at the center of the biggest technological shift in decades…

And the stock drops over 30%.

Most investors saw that and assumed one thing:

“AI is overhyped. The spending is getting out of control.”

That’s the easy take.

But the real story sits somewhere most people aren’t even looking.

And once you see it, the entire sell-off starts to feel less like a warning… and more like a setup.

Upgrade to paid

The Market Is Focused On The Wrong Problem

Right now, the narrative around Microsoft is simple:

• Capital expenditure is exploding

• Data centers are expensive

• GPUs aren’t cheap

• Energy costs are rising

• Azure growth isn’t “fast enough”

So the conclusion?

Microsoft is spending too much… too early… with uncertain returns.

That’s what pushed the stock down roughly 35% from its highs.

Upgrade to paid

But here’s the mistake.

Investors are focusing on when revenue shows up…

Instead of asking whether that revenue already exists.

The $625 Billion Clue Sitting In Plain Sight

Buried inside Microsoft’s earnings is a number most people completely ignored.

$625 billion.

That’s Microsoft’s commercial remaining performance obligations (RPO).

In simple terms:

This is contracted revenue that hasn’t been recognized yet.

It’s not speculation.
It’s not projections.
It’s not “maybe.”

It’s signed business.

Waiting to be delivered.

Now think about what that means.

While the market is worried that Microsoft is building too much AI infrastructure…

Microsoft already has hundreds of billions in demand lined up.

This flips the entire narrative.

The question isn’t:

“Is Microsoft overbuilding?”

It becomes:

“Is Microsoft building fast enough to meet demand?”

Upgrade to paid

Azure Isn’t Slowing — Expectations Were Just Unrealistic

Another piece of the sell-off came from Azure.

Growth came in strong… but not strong enough for a market that had priced in perfection.

Azure is already running at over $75 billion annually.

And it’s still growing close to 40%.

At that scale, those numbers are massive

But here’s what matters more.

Azure is the backbone of Microsoft’s AI strategy.

Every Copilot interaction… every enterprise AI deployment… every model running in production…

Flows through Azure.

So when AI demand rises, Azure doesn’t just benefit.

It becomes the toll road.

And Microsoft owns it.

Upgrade to paid

The OpenAI Relationship: Power + Risk

This is where things get more interesting.

Roughly 45% of Microsoft’s commercial backlog is tied to OpenAI-related workloads.

On top of that, OpenAI has committed around $250 billion in Azure services.

That’s massive.

It means a huge portion of Microsoft’s future revenue is directly linked to AI adoption.

But it also introduces a real risk.

Reports suggest tension between Microsoft and OpenAI after OpenAI explored partnerships with Amazon.

At the core, it’s about control.

Who gets to host the most valuable AI workloads?

Microsoft wants exclusivity.

OpenAI wants flexibility.

This isn’t just a partnership issue.

It’s a battle over where the future of AI infrastructure lives.

Still, both sides depend on each other.

Microsoft provides the infrastructure and distribution.

OpenAI drives demand and usage.

So the most likely outcome isn’t a breakup.

It’s renegotiation.

And if handled right, that could actually expand the opportunity instead of limiting it.

AI Is Already Showing Up In Real Usage

This isn’t theoretical anymore.

Copilot has crossed:

• 4.7 million paid users

• 10x growth in daily activity

• ~900 million monthly reach across AI features

That’s not early-stage testing.

That’s adoption.

And adoption is what converts infrastructure spending into revenue.

Which is exactly what investors claim they’re waiting for.

Upgrade to paid

The Valuation Reset No One Is Talking About

After the drop, Microsoft trades around:

• ~22x forward earnings

• PEG ratio ~1.39

This is the cheapest the stock has been in years.

Not because the business got worse.

But because expectations got pulled forward too aggressively… and are now being reset.

This matters.

Because great companies rarely go on sale when everything looks perfect.

They go on sale when the narrative breaks.

What The Market Is Missing

Put everything together and a pattern starts to form:

• Massive AI spending → visible

• Temporary margin pressure → visible

• Slower than expected Azure growth → visible

But at the same time:

• $625B in contracted demand → ignored

• Explosive AI usage → underappreciated

• Deep integration across enterprise software → misunderstood

The market is reacting to what’s happening today.

But Microsoft is building for what’s already committed tomorrow.

Upgrade to paid

How To Actually Play This As An Investor

This is where most people get it wrong.

They try to time the bottom.

They wait for the “perfect” entry.

They want confirmation.

By the time that comes… the opportunity is gone.

A smarter approach is simpler.

Think in layers.

1. Start building a position at current levels

2. Add on further weakness

3. Focus on long-term demand, not short-term noise

Because if the backlog converts the way it should…

The current price won’t matter much in a few years.

The Real Question

This isn’t about whether Microsoft is a good company.

That part is obvious.

The real question is:

Do you believe AI demand is real… and already locked in?

Because if it is…

Then this sell-off isn’t a warning.

It’s a mispricing.

And those don’t last long.

If you want more breakdowns like this — where the focus is on what the market is missing, not what it’s already reacting to — that’s exactly what gets shared inside The Daily Dollar.

That’s where the real edge is built.

Subscribe now

Archer Aviation has become one of those stocks where the story feels bigger than the numbers. On the surface, it’s easy to get pulled in. A new category, futuristic aircraft, big partnerships, and a promise that people could move across cities in minutes instead of sitting in traffic.

But when you step back, the market isn’t rewarding the story right now. It’s questioning it.

And that shift matters more than most investors realize.

The Core Question No One Can Ignore

Every speculative company eventually faces the same test: does execution match the narrative?

With Archer Aviation, that question has moved to the front of the conversation.

The company is building electric vertical takeoff and landing aircraft, commonly known as eVTOLs. These aircraft are designed to operate as air taxis, moving passengers across cities using short, battery powered flights. On paper, it’s a compelling vision.

But vision alone doesn’t move markets.

Execution does.

Read more

Subscribe now

Everyone is calling IREN an AI company now.

That’s the headline. That’s what’s getting priced in.

But if you actually look under the hood, the story is more complicated. It’s still a Bitcoin mining company funding an aggressive shift into AI infrastructure. And that gap between narrative and reality is where everything comes together.

Not just upside.

Also risk.

This is not about hype. It’s about understanding what has to go right for IREN to justify what the market is already expecting.

The Core Setup: Power First, Revenue Later

Let’s start with the number that gets attention:

• 4.5 gigawatts of secured power capacity

On paper, that scale points toward massive revenue potential. At full utilization, estimates suggest something in the range of $30B+ annual revenue potential.

But here’s the part that actually matters:

• Only ~460 MW is active

• 2.1 GW is under construction

• 1.6 GW is still in development

So less than 10% of the capacity is actually producing revenue today.

Everything else?

Still being built.

This is why the market is so divided.

Some investors see a future AI infrastructure giant.

Others see a company still trying to finish its buildout while carrying high expectations.

Both views can be true at the same time.

Read more

Firefly Aerospace (FLY) has moved beyond its early identity as a speculative launch startup. It is steadily becoming a vertically integrated defense technology company with a meaningful software component.

The market has not adjusted to that shift.

Attention remains stuck on short-term issues and the lingering weight of class-action litigation. That gap between perception and reality is where the opportunity sits.

Read more

At first glance, this company does not inspire confidence. A $354 million negative free cash flow number in 2025 is the kind of figure that makes most investors stop reading and move on.

That reaction is understandable, but it may also be missing the point.

Read more

The Silver Economy is not the easy demographic tailwind traditional healthcare investors believe it is. In many ways it is the opposite. What looks like a guaranteed demand boom for assisted living real estate may actually become the moment that breaks the industry. At the same time it could create one of the biggest growth waves yet for consumer robotics and home health technology.

Read more

Subscribe now

Bottom Line Up Front (BLUF): Global copper markets have entered a severe, irreversible structural deficit. Driven by the price-inelastic “triple threat” of AI data centers, grid modernization, and global electrification, demand is outstripping a crippled supply chain. With treatment and refining charges (TC/RCs) crashing to zero and mega-miners choosing to merge rather than build, the recent pullback to $5.80/lb ($13,100/ton) offers a rare, high-conviction entry point before prices structurally rebase toward $15,000/ton.

1. Retail investors are playing the AI supercycle wrong.

Retail investors are blindly throwing capital at software companies trading at 40x sales. They are obsessing over the latest language models, cloud revenue multiples, and semiconductor yields. They are missing the forest for the trees.

You cannot build a gigawatt AI data center out of thin air. You cannot modernize a crumbling electrical grid with software. You need the oldest, dirtiest industry on earth. You need copper.

Right now, a massive physical bottleneck is forming. The world has run out of easy copper, and the timeline to permit and build a new tier-one mine is stretching past 15 years. The market is attempting to price a futuristic, hyper-electrified world using a supply chain that hasn’t seen meaningful capacity expansion in a decade.

This is no longer a cyclical industrial metal trade. This is a geopolitical hostage situation. The structural deficit is here, and the window to front-run institutional capital is closing.

2. The Big Picture (Macro + Market Context)

To understand the opportunity, you must understand the exact macro regime we are trading in as of March 2026.

The broader US market remains heavily concentrated in mega-cap tech, leaving traditional value and cyclical sectors historically under-allocated. While the S&P 500 grapples with lofty valuations and a shifting interest rate environment, underneath the surface, a brutal reality is setting in for the physical economy.

In January 2026, copper prices violently spiked to an all-time high of roughly $14,500 per metric ton (over $6.50/lb). What followed was a classic, healthy technical correction. Driven by a surging US dollar, a temporary spike in LME warehouse inventories, and hedge funds unwinding speculative long positions, the price has consolidated back to the $13,100/ton ($5.80/lb) range.

Retail investors see the red on the screen and assume the “commodity supercycle” narrative is dead. They are completely overlooking institutional behavior.

Just look at the recent $60 billion Anglo American and Teck Resources merger. When the smartest operators in the mining industry decide to spend tens of billions to buy existing copper assets rather than build new ones, they are sending you a glaring signal: the permitting wall is impenetrable, and new supply is not coming online fast enough to save us.

Subscribe now

3. Deep but Clear Analysis

This trade is built on three unshakeable pillars: collapsing supply, inelastic demand, and broken market mechanics.

The Fundamentals: A Historic Shortage We are entering what Morgan Stanley is calling the most severe supply deficit in 22 years, projecting a shortfall of 590,000 tonnes in 2026 that will aggressively widen to 1.1 million tonnes by 2029.

Why? Because demand is compounding across multiple vectors simultaneously. The energy transition alone requires millions of tonnes of new wire and cable. But the true wildcard is Artificial Intelligence. J.P. Morgan estimates that in 2026 alone, data center installations will consume 475,000 metric tons of copper. AI servers require exponentially more power distribution and liquid cooling infrastructure than traditional cloud computing. This demand is entirely price-inelastic; Microsoft and Amazon will not pause a $2 billion data center build just because copper prices rose 20%.

The Institutional Signal: Smelter Fees at Zero If you want to know how tight the market is, look at the midstream operators. Copper smelters charge miners a fee, known as Treatment and Refining Charges (TC/RCs), to process raw ore into refined metal. In early 2026, the annual TC benchmark effectively crashed to $0 per tonne.

This is unprecedented. It means smelters are so desperate for raw copper concentrate to keep their massive facilities running that they are willing to process the ore for free. There is simply not enough physical rock coming out of the ground to feed the system.

Technical Structure Technically, copper is in a long-term structural uptrend characterized by higher highs and higher lows. The massive breakout in early 2026 cleared decades of overhead resistance. The current pullback to the $5.80/lb region represents a critical retest of the upper Bollinger Bands on the monthly chart. Momentum indicators (like the ADX) remain highly elevated, confirming that this is a trend expansion phase, not a mean-reversion market.

The Substitution Myth The primary bear argument is that high copper prices will force manufacturers to substitute with aluminum. This is a mathematical blind spot. In 2023, net substitution accounted for roughly 1.5% of total copper use. Even factoring in material miniaturization, the loss is a rounding error compared to the 11% compound annual growth rate in green wire and cable demand. The sheer volume of new demand effortlessly absorbs these marginal substitution losses.

Subscribe now

4. What Long-Term Investors Should Do

For the long-term portfolio builder (3 to 5-year horizon), the mandate is clear.

• Clear Stance: Accumulate.

• Ideal Strategy: Staged buying. Do not blow your entire cash position today. Use this current consolidation phase to scale into a core position.

• Portfolio Allocation: 5% to 8% of your total portfolio, depending on your risk tolerance and existing exposure to the broader energy/materials sector.

• How to Play It:

  • The Anchor: BHP Group (BHP) or Freeport-McMoRan (FCX). These are the mega-caps. They possess the fortress balance sheets required to weather the storm and pay dividends while you wait for the thesis to play out.

  • The Growth Tilt: Ivanhoe Mines (IVN.TO). Operating massive, high-grade discoveries in the DRC, they offer significantly higher beta to the rising spot price of copper.

  • The Direct Play: Sprott Physical Copper Trust (COP.U). If you want to avoid individual mining risk (jurisdiction risk, management execution, flooding), buy the physical trust. It tracks the raw asset.

What Confirms the Thesis: Continued M&A activity in the mining sector, rising premiums for physical delivery, and sustained AI capital expenditure from Big Tech. What Invalidates the Thesis: A sudden, global depression that permanently destroys industrial demand, or an unforeseen technological breakthrough in room-temperature superconductors (highly unlikely in our time horizon).

Subscribe now

Chart Analysis: The Anatomy of a Squeeze

• The Parabolic Breakout: Notice the violent, near-vertical surge into early 2026. Decades of structural resistance were obliterated on explosive volume as the physical deficit became undeniable.

• The Healthy Digestion: We are currently witnessing a textbook consolidation. The market is cooling off extreme momentum and shaking out weak retail hands, all while holding firmly above key moving averages.

• The Institutional Floor: Look at the volume profile on the right axis. The massive accumulation nodes (thick blue/pink bands) confirm that institutional capital has built an ironclad floor of support directly beneath us.

The verdict: This isn’t a macro top. This is a launchpad.

5. What Swing Traders Should Do

For those looking to capture the next 3 to 6-month rotation, copper offers a highly asymmetrical setup right now.

• Current Technical Bias: Bullish long-term; Neutral-to-Bullish short-term (consolidation phase).

• Key Price Levels (Spot Copper):

  • Support: Strong fundamental and technical support sits around $5.10/lb. A deeper, “line in the sand” support sits at $4.40/lb, though it is highly unlikely we see that without a systemic market shock.

  • Resistance: $6.20/lb is the immediate hurdle, followed by the psychological and structural ceiling at $6.50/lb (the recent all-time highs).

• The Setup: * If price holds above $5.70/lb on the weekly close... initiate a half-sized long position in a high-beta vehicle like the Global X Copper Miners ETF (COPX).

  • If price breaks out and closes above $6.20/lb on heavy volume... add the remaining half of your position, as this signals the end of the consolidation phase and the start of the next leg up toward $7.00/lb.

• Stop-Loss Logic: Place a hard stop below $5.00/lb on a daily closing basis. If copper loses the $5 handle, the structural deficit narrative has likely been temporarily overwhelmed by a severe macroeconomic liquidity event, and you do not want to hold the bag.

• Catalyst Watchlist: Watch the upcoming US policy announcements regarding potential 15% to 25% tariffs on refined copper. If tariffs are announced, expect a massive short-term spike as importers frantically stockpile metal before implementation.

  Subscribe now

Chart Analysis: X-Raying the Order Flow

• Peeling Back the Tape: This isn’t a standard price chart—it’s a volume footprint. It strips away the noise and shows exactly where the money is transacting inside the candlesticks.

• Absorbing the Panic: Look at the dense transaction clusters on this recent pullback. The heavy red volume blocks represent aggressive selling, but notice how the price refuses to collapse. That is “smart money” passively absorbing the panic with massive limit orders.

• The Accumulation Footprint: Down at the bottom, the Open Interest and Delta data confirm the story. The selling pressure is drying up as it hits structural support. The “shakeout” is nearing completion.

The verdict: Weak hands are panic-selling the dip. Institutional capital is quietly buying every single ounce of it.

6. Risk Management Framework

You cannot trade institutional-grade setups with retail-grade risk management.

• Position Sizing: Copper mining stocks carry inherent operational risks (strikes, grade degradation, political nationalization). Never allocate more than 2% of your total capital to a single junior or mid-tier mining stock. Diversify through ETFs or a basket of mega-caps.

• Correlation Awareness: Copper is highly sensitive to the US Dollar Index (DXY). Because the metal is priced in dollars globally, a surging DXY makes copper more expensive for foreign buyers, typically suppressing the price. If the Federal Reserve unexpectedly turns hawkish and the dollar rips higher, copper equities will take a temporary beating. Monitor the DXY closely.

• China Exposure: Despite the shift toward India and AI, China still consumes roughly half of the world’s copper. Weakness in the Chinese property sector is currently baked into the price, but a sudden collapse in their manufacturing sector would introduce severe volatility.

  ---

7. Scenario Planning

Professional strategists do not predict the future; they prepare for it. Here is how this plays out over the next 12 to 18 months.

The Bull Case (30% Probability): The Squeeze Materializes Supply disruptions worsen (e.g., strikes in Chile, export bans in Indonesia). The US administration formally announces a 25% tariff on refined copper, sparking panic hoarding. AI data center build-outs accelerate past Wall Street estimates. What to do: Hold full positions. Move stop-losses up to breakeven, then trail them tightly behind major moving averages. Take initial profits on your swing trades when spot copper crosses $6.75/lb.

The Base Case (55% Probability): The Steady Grind The market digests the January highs. Supply remains stagnant, but macroeconomic headwinds (stubborn inflation, slower global growth) prevent a parabolic melt-up. Copper grinds steadily higher, averaging $11,500 to $12,000 per ton as the physical deficit slowly bleeds the LME warehouses dry. What to do: Continue dollar-cost averaging into your core long-term positions. Sell out-of-the-money covered calls against your mining stocks to generate yield during the slow, choppy grind upward.

The Bear Case (15% Probability): The Macro Break A severe global recession takes hold. Manufacturing PMIs globally collapse, and the highly anticipated AI infrastructure spending is slashed as Big Tech focuses on margin preservation. The structural deficit is masked by a cyclical collapse in demand. Copper falls back below $4.50/lb. What to do: Execute your hard stops on all swing trades. For long-term portfolios, do not panic sell at the bottom. The geological deficit has not changed; it has only been delayed. Use the severe weakness to aggressively upgrade the quality of your portfolio, swapping out of risky junior miners into mega-caps at generational discounts.

Subscribe now

The smartest money in the world is quietly buying up the physical infrastructure of the future. Make sure you own a piece of it before they lock the gates.

Most investors completely ignore 13F season.

They scroll past it.
They assume it’s backward looking.
They think the opportunity is already gone.

That’s exactly why it matters.

Because while retail investors argue about headlines and price charts…

The biggest capital allocators on earth just made some of the most aggressive position shifts we’ve seen in years.

Massive allocations.
Double-digit portfolio weights.
Infrastructure bets measured in hundreds of billions.

And buried inside those filings is one clear pattern.

A pattern that tells you where long-term capital believes the next decade of returns will come from.

This isn’t about copying trades.
This is about understanding how serious money thinks.

In this breakdown, we’re going to go deeper than the headlines:

• Why Seth Klarman made Amazon his #2 position overnight
• Why Bill Ackman is building a concentrated AI barbell
• Why Berkshire cut Apple again — and why that matters more than people think
• Why Meta at 22x earnings may not be priced the way the market assumes
• What the capex explosion across Big Tech actually implies for future margins
• And the quiet structural shift happening at Berkshire Hathaway

Most people will glance at the filings.

You’re going to study them.

Why 13Fs Still Matter (If You Know How to Read Them)

A 13F filing shows U.S. equity holdings for managers with over $100 million in assets.

Yes, it’s delayed 45 days.
Yes, it excludes derivatives and international holdings.

But here’s what most investors miss:

13Fs don’t just show what was bought.
They show conviction.

Portfolio weight reveals belief.
Position size reveals risk tolerance.
Timing reveals strategic intent.

And Q4 2025 revealed something unusual:

Large, disciplined investors are leaning aggressively into AI infrastructure — but not in the way most people think.

Let’s break it down.

Read more

Alphabet delivered one of the strongest operating years in its history.

Revenue beat.
Earnings beat.
Cloud reaccelerated.
Margins expanded.
Backlog exploded higher.

And the stock fell 9%.

When a company puts up numbers like this and sells off, serious investors don’t panic.

They investigate.

I spent the past week dissecting Alphabet’s Q4 and full year 2025 earnings from every angle: segment growth, margin structure, operating leverage, backlog quality, capex deployment, AI positioning, valuation compression risk, and forward cash flow inflection points.

What I found is not a business slowing down.

It’s a business aggressively repositioning itself to become the infrastructure backbone of AI.

Let’s walk through this properly.

This is not a surface-level recap.

This is what actually matters.

Read more

Subscribe now

Something doesn’t add up.

Amazon just reported:

• $716.9 billion in full-year revenue

• 24% AWS growth

• Gross margins crossing 50% for the first time in company history

• Net income up 31% year over year

  

  Subscribe now

  

And yet… the stock fell about 10% after earnings and now sits roughly 11% below recent highs.

So what is the market seeing that the headlines are missing?

Over the last few days, I went through the full earnings release, segment breakdowns, and a new regulatory ruling out of Germany. What I found isn’t obvious at first glance.

This isn’t just about whether Amazon beat estimates.

This is about whether the quality of Amazon’s revenue is changing in a way that the market hasn’t fully priced in yet — and whether one European decision could quietly become a bigger issue.

Let’s break this down properly.

Subscribe now

The Real Numbers Behind Q4 and Full-Year 2025

On February 5, 2026, Amazon reported fourth-quarter results that technically “beat.”

Q4 2025

• Revenue: $213.4 billion (up 14% year over year)

• AWS revenue: $35.6 billion (up 24%)

• Gross profit: $103.4 billion (up 16%)

• Gross margin: 48.47% (expanded 1.3 percentage points)

• Net income: $21.2 billion

That margin expansion is not a small detail.

For decades, Amazon was known as a low-margin, high-volume machine. Retail drove the top line. AWS carried profitability. Everything else was secondary.

Now gross margin is pushing 50%.

That tells you something fundamental is shifting.

Full-Year 2025

• Revenue: $716.9 billion (up 12%)

• Gross profit: $316.5 billion (up 16%)

• Gross margin: 50.29%

• Operating income: $80 billion (up 17%)

• Net income: $77.7 billion (up 31%)

• Diluted EPS: $7.29

A few years ago, the idea that Amazon would keep more than half its revenue after direct costs would have sounded unrealistic.

Now it’s happening.

So why isn’t the stock exploding higher?

Subscribe now

Why the Market Isn’t Rewarding It (Yet)

Amazon currently trades around 26x forward earnings — roughly 40% above the sector median.

That multiple isn’t cheap.

But here’s what’s really holding sentiment back:

1. Massive AI Capex

Management plans to spend $200 billion in 2026 on AI infrastructure and data centers — the highest projected spend among the Magnificent Seven.

Investors are worried about:

• Return on capital

• Timing of monetization

• Whether cloud growth re-accelerates enough to justify it

Heavy spending without immediate profit expansion makes Wall Street uncomfortable.

2. Narrative Confusion

AWS isn’t growing at 40% anymore.

Retail still carries a “low margin” stigma.

Advertising and AI services are growing fast — but analysts haven’t fully rebuilt their valuation models around them.

So Amazon sits in a strange position:
Too diversified for one dominant story.
Too early in the next chapter for consensus excitement.

And sometimes that’s exactly where opportunity lives.

Subscribe now

Growth Driver #1: Alexa Plus (The Sleeper Bet)

On February 4, 2026, Amazon launched Alexa Plus.

• $19.99 per month standalone

• Free with Prime

  

There are over 600 million Alexa-enabled devices globally.

Subscribe now

Let’s do the math.

If just 5% convert to paid:

• ~$1.8 billion quarterly revenue

• ~$7.2 billion annually

At 10% conversion:

• ~$14.4 billion annually

This would come from a division historically considered stagnant.

Early metrics are interesting:

• 2x conversations

• 3x purchases

• 5x recipe requests

If Amazon begins reporting Alexa Plus subscriber numbers, that will be the signal this is becoming meaningful.

Growth Driver #2: Advertising (The Quiet Giant)

This is where I think the market is underestimating things most.

Amazon’s ad business generated:

• $21.3 billion in Q4 revenue

• 23% year-over-year growth

  Subscribe now

  

That is no longer a side hustle.

Through its demand-side platform, Amazon now allows advertisers to buy inventory on platforms like Netflix and Spotify — while leveraging Amazon’s first-party data.

That data advantage matters.

Add to that:

• AI-powered campaign tools

• Full-funnel ad capabilities

• Prime Video’s growing ad-supported audience (up 58% since April 2024)

  Subscribe now

  

Even Thursday Night Football is pulling in over 15 million average viewers per broadcast, with playoff games hitting 31.6 million streams.

This is turning Prime Video into a real ad engine.

Five years ago, AWS carried profitability.

Today, advertising is becoming a second profit pillar.

And markets rarely price in quiet transformations early.

The Regulatory Risk That Could Spread

Now we get to the part long-term investors can’t ignore.

On the same day as earnings, Germany’s Federal Cartel Office ruled against Amazon’s algorithmic price controls.

The regulator banned Amazon from automatically flagging or removing third-party listings based on algorithmic pricing decisions that weren’t transparent to sellers.

Amazon was also ordered to repay €59 million in profits related to those practices.

Financially?

That amount is irrelevant to Amazon’s scale.

Strategically?

It’s more important.

Germany is Amazon’s second-largest European market, controlling roughly 60% of online retail sales there.

And this ruling was coordinated with the European Commission.

If similar reasoning is adopted under the EU’s Digital Markets Act, these restrictions could expand across Europe.

Subscribe now

Amazon has:

• One month to appeal

• Three months to propose compliance

• Nine months to fully implement changes

The risk isn’t the fine.

The risk is precedent.

Regulation that reshapes how Amazon controls pricing and marketplace visibility could compress margins over time if expanded broadly.

It’s manageable — but it’s real.

So Is This a Buying Opportunity?

Here’s where I land.

The market is focused on:

• AI spending

• Slowing AWS growth vs historical highs

• Regulatory noise

What it may be underpricing:

• A structural shift toward higher-margin revenue

• A rapidly scaling ad business

• A potential subscription AI layer in Alexa Plus

• Gross margins now crossing 50%

The quality of revenue is changing.

Subscribe now

Retail is no longer the entire story.
AWS is no longer the only profit engine.
Advertising is maturing.
AI services are expanding.
Subscriptions are deepening.

At 26x forward earnings, Amazon is not cheap.

But when you look 3 to 5 years out, you’re buying:

• E-commerce scale

• Logistics infrastructure

• Cloud dominance

• Advertising growth

• AI monetization

• Global distribution power

All inside one company.

Short-term volatility? Absolutely.

Structural decline? I don’t see it.

The German ruling is worth monitoring. Heavy capex will pressure margins near term. But Amazon’s balance sheet and cash generation give it room to execute.

Sometimes the best entries don’t come when everything feels exciting.

Subscribe now

They come when the narrative is unclear — but the numbers are improving.

The stock is down.
Margins are expanding.
Revenue mix is improving.
New monetization layers are forming.

That tension is where long-term returns are born.

Now the question becomes simple:

Are investors willing to look past the next quarter… and focus on the next five years?

When Geopolitics Becomes the Market

Markets can price inflation.
They can price earnings.
They struggle to price regime collapse.

The latest escalation between the United States and Iran marks a structural shift from contained retaliation to direct leadership targeting. That changes the calculus. This is no longer a “limited engagement” narrative. It is a regime stability event.

And regime stability events don’t move in percentages. They move in shockwaves.

The immediate consequence is a repricing of geopolitical risk across energy, defense, logistics, currencies, and volatility products. The secondary consequence is a liquidity response from global central banks. Investors who treat this as headline noise are underestimating the structural shift underway.

The Macro Context: The Strait of Hormuz and the 20% Rule

The most important variable is not politics.

It is oil flow.

The Strait of Hormuz handles roughly 20% of global oil supply. Even a temporary disruption introduces a supply shock large enough to:

• Spike crude prices sharply

• Lift inflation expectations

• Delay Federal Reserve easing

• Strengthen the U.S. dollar initially

• Tighten global financial conditions

Energy markets transmit instantly. Supply chains follow.

If Gulf airspace closures expand or maritime insurance rates surge, global logistics costs rise. That hits airlines, shipping firms, European manufacturers, and emerging markets dependent on imported fuel.

This is how a regional conflict becomes a global macro event.

The Federal Reserve’s path becomes hostage to oil volatility. A rate cut cycle during an energy shock becomes politically and economically difficult.

Core Sector Analysis: Where Capital Rotates

Markets are already bifurcating. Some sectors gain structural tailwinds. Others face compression.

Read more