Apple just reported $111.2 billion in quarterly revenue, up 17%. Earnings beat. iPhone up 20%. Services at an all-time high. The stock popped 4% after-hours.

That’s the headline. But the real story is buried two lines down in the cash flow statement.

Apple cut its capex from $6 billion to $4.3 billion.

Down 28%, in the same quarter Microsoft, Google, Meta and Amazon collectively guided to roughly $700 billion of 2026 AI capex. Down in the quarter that John Ternus was named to take over as CEO on September 1. Down at exactly the moment every other large-cap tech company is throwing the kitchen sink at GPUs and data centers.

Either Apple is wildly behind. Or Apple is making the most contrarian bet in big tech right now.

If you’re trying to figure out where wealth gets created in the next decade of AI, this is the question to sit with.

The capex chasm just got wider

Two days ago we wrote that Mag 4 — Microsoft, Alphabet, Meta and Amazon — confirmed combined 2026 AI capex of around $700 billion. Bigger than Switzerland’s GDP. Bigger than every dollar of venture capital deployed in any year on record.

Apple’s quarterly capex run-rate, annualised, comes out to roughly $17 billion. That’s not 1/10th of what the hyperscalers are spending. It’s closer to 1/40th.

Inside the same Magnificent 7, you now have two completely different business models cohabitating under one ETF ticker.

The capex bulls — MSFT, GOOGL, META, AMZN, NVDA — are betting that owning the compute layer is the moat. They’re spending hundreds of billions a year to make sure no upstart can rent their way to scale.
The capex bear — AAPL — is betting that the device, the OS, and the user relationship is the moat. They’ll let Google power the foundation model (Apple Intelligence already leans on Gemini) and pour their incremental dollars into R&D and on-device silicon instead.

Both bets can win. They cannot both win to the same degree. And right now the market is pricing them as if they’re the same trade.

R&D up 33%, capex down 28%

This is the number every investor missed in the headlines. R&D in Q2 jumped to $11.4 billion, up 33% year over year. Capex fell 28%.

Apple is spending more than ever on AI. They’re just spending it on people and silicon design, not on warehouses full of NVIDIA chips.

That distinction is everything.

When Microsoft commits $80 billion to data centers, that money turns into depreciation on the balance sheet for the next decade. It’s an asset, but it’s a heavy one. If GPU prices fall 50% in three years (and they will — see yesterday’s piece on Mag 4 earnings), every hyperscaler is sitting on $50–100 billion of impaired equipment.

When Apple spends $11.4 billion on R&D, that money turns into the M-series and A-series chips, the Neural Engine, the on-device AI stack. The asset is human capital and IP, not a building full of cards that depreciate the moment a new generation ships.

In a world where compute gets dramatically cheaper every 18 months — which is the world we’re in — being asset-light is a feature, not a bug.

The Ternus question

September 1, John Ternus takes over from Tim Cook as CEO. Ternus is a hardware guy. He ran the iPhone, iPad, Mac and Apple Silicon teams. He spearheaded the launch of the MacBook Neo this March — Apple’s first low-cost MacBook in a generation.

His mandate from the board, by every report we’ve seen, is one thing: fix Apple’s AI strategy.

What does “fix” mean if you’re a hardware-first CEO? It almost certainly does not mean spending $80 billion a year on data centers. It means tighter integration. On-device inference. Models that run on the Neural Engine instead of in the cloud. Foundation models licensed from whoever has the best one — for now, Google’s Gemini — until Apple’s own internal model catches up or until open-source eats the foundation model layer entirely.

If Ternus is right, the AI economy has two profit pools:

The compute layer — captured by NVIDIA and the hyperscalers who own the data centers.
The device and distribution layer — captured by whoever owns the relationship with the end user.

Apple is the only Mag 7 company aiming exclusively at #2. Every other Mag 7 company is fighting for a piece of #1.

The contrarian wealth play

Here’s where this matters for your portfolio.

The consensus AI trade right now is “buy the basket” — own NVDA, MSFT, GOOGL, META, AMZN, and pretend they’re one position. That trade has been broken for a week and a half (see Alphabet +6% vs Meta -5% on the same earnings night).

Apple’s print just made the basket completely incoherent.

You can’t own AAPL and MSFT for the same reason. They are now opposite bets.

If you believe the foundation-model wars get bloodier and only the biggest spenders survive — buy MSFT, GOOGL, META. Skip AAPL.
If you believe the foundation model gets commoditized and the asset-light players capture the consumer surplus — buy AAPL. Be skeptical of the capex bulls.
If you don’t know — at least know that owning all of them at once is not a hedge. It’s a confused thesis.

The interesting wealth lesson is what should be obvious by now: when an entire sector trades together, the contrarian setup is when one component breaks the correlation. Apple just broke the correlation. The market hasn’t fully repriced yet.

What to actually do this week

Three concrete moves to consider — not advice, just framework:

Pick a side. Decide whether your AI thesis is “compute infrastructure” or “device and distribution.” Then look at your portfolio honestly. If you own all of Mag 7 in equal weight, you don’t have a thesis. You have a label fund.
Track your concentration. When Apple alone is 7%+ of the S&P 500 and Mag 7 collectively is roughly 30%, “the index” is a single concentrated AI bet whether you like it or not. Know your real exposure.
Watch services and on-device AI rollout. Apple’s services revenue at all-time highs is the early signal that the asset-light strategy is working at the unit-economics level. If on-device AI starts to convert into incremental services revenue, the thesis is right.

If you’re in the early innings of building wealth, what matters more than picking the right Mag 7 bet is having a thesis. The capex chasm is a reminder that even the most boring index fund is now an opinionated portfolio. You may as well form your own opinion.

Want to track your own path to financial independence? The Wealtharian Wealth Tracker lets you monitor your net worth, FU money progress, and investment milestones in one place. Try it free →