Apple Inc. (AAPL  ) is fighting the AI race with engineers and code rather than steel and silicon. Through the first nine months of fiscal 2026, Apple spent $34.04 billion on research and development, according to the company's financial statements. Over the same stretch, capital expenditures - mostly payments for property, plants and equipment - totaled just $6.8 billion. That means Apple's R&D spending ran roughly five times its capex, a ratio unlike anything else in Big Tech's AI buildout.

Microsoft's GPU-Heavy Capex Bet

Compare Apple to Microsoft Corp. (MSFT  ). In its fiscal third quarter alone, Microsoft's capital expenditures totaled $31.9 billion, with roughly two-thirds of that allocated to short-lived assets such as GPUs and CPUs.

At the time, CFO Amy Hood guided calendar-2026 capex to roughly $190 billion. Microsoft has since revised that figure down to about $175 billion, though the change reflects an accounting shift - extending the useful life of data centers and office buildings from 15 to 25 years, which moves some leases off the capex ledger - not a pullback in actual AI investment.

Meta Raises Its Capex Outlook

Meta Platforms Inc. (META  ) is running an even steeper curve. The company spent $31.08 billion on capex, including finance lease payments, in its second quarter of 2026, according to its earnings release. Management has lifted its full-year 2026 capex outlook to $130 billion-$145 billion, up from a prior $125 billion-$145 billion range.

Two Different Bets on AI

The contrast captures two very different theories of how to win the AI era.

Microsoft and Meta are racing to lock down compute - betting that owning enough GPUs, servers and data center capacity secures a durable edge in training and running frontier models, even if it means single quarters of spending that dwarf Apple's entire nine-month capital budget.

Apple is making the opposite bet: that AI's real value to a consumer hardware company gets captured in silicon design, software integration and the products layered on top of a 2-billion-plus-device installed base, not in owning the data centers underneath it.

It's a wager with real risk on both sides. If frontier AI capability keeps compounding, Apple's lighter infrastructure footprint could leave it dependent on outside partners for the heaviest compute.

If capacity outpaces demand, Microsoft and Meta absorb the depreciation on hundreds of billions in hardware. For now, the numbers show three of tech's biggest AI spenders pulling in genuinely different directions.