The More Artificial Intelligence (AI) Models That Come Out, the More I'm Convinced Apple Has the Right Strategy

Apple is projected to spend just $14 billion on capex in 2026, far below peers like Meta, which may spend up to $135 billion on AI. The article argues Apple’s slower, more selective AI strategy may still be effective because AI development costs could fall over time and its ecosystem reduces customer churn. Overall, the piece is a bullish long-term case for Apple, but it is mostly commentary rather than a near-term catalyst.

Analysis

Apple’s low-capex stance is less a weakness than a capital-allocation signal: it is implicitly betting that frontier AI will commoditize faster than the market expects, which would compress the advantage of brute-force model spending. If that happens, the incremental return on AI dollars spent by hyperscalers likely decays sharply over the next 12-24 months, and the value migrates from model training to distribution, device integration, and on-device inference. That is structurally favorable to Apple’s installed base and to suppliers that can monetize edge compute without participating in the full capex arms race.

The second-order winner is not just AAPL; it is any company selling picks-and-shovels for inference efficiency rather than training scale. If AI feature costs fall, the market may rotate away from “who can spend the most” toward “who can ship the best user experience at the lowest marginal cost,” which supports premium hardware ecosystems and weakens the ROI case for the most aggressive AI spenders. META is the clearest near-term beneficiary of this narrative if investors continue to reward rapid model iteration, but the longer-duration winner is the platform that can amortize AI over billions of devices without materially lifting its cost base.

The key risk is timing: Apple can underinvest for a while, but if AI becomes a visible consumer feature set over the next 6-9 months and Apple’s products lag meaningfully, sentiment can deteriorate fast even if fundamentals remain intact. A sharper-than-expected AI UX leap from competitors would also raise the switching narrative at the margin, particularly in premium smartphones where upgrade cycles are already stretched. Conversely, any credible proof that Apple can deliver on-device AI with minimal power and privacy tradeoffs would re-rate the stock by reframing its “slowness” as discipline.

Consensus is likely overestimating the importance of capex intensity and underestimating the value of distribution, retention, and device-level lock-in. The market keeps treating AI as a race to the largest model, but if distillation and inference efficiency keep improving, the economic moat shifts to ecosystem control. That makes this a quality-vs-growth trade more than an AI trade.

Key Decisions for Investors

• Hold AAPL as a long-duration quality compounder; add on 5%-8% pullbacks over the next 1-3 months if the market sells the stock on perceived AI underinvestment, since the core risk/reward is that optionality gets mispriced lower than the downside to the ecosystem.
• Pair trade: long AAPL / short a basket of high-AI-capex names (META as the cleaner single-name hedge if basket execution is undesirable) for 3-6 months, betting that investors rotate from spend-intensity to free-cash-flow durability as AI margins normalize.
• Buy medium-dated AAPL call spreads or bull call spreads into any post-earnings volatility event over the next 1-2 quarters; the best setup is if the market extrapolates slow AI rollout too far, because even modest product proof points can drive multiple expansion.
• Avoid chasing pure AI capex winners at peak enthusiasm; if model costs continue to fall, the incremental payoff to training spend can compress quickly, creating a left-tail risk for names priced on perpetual infrastructure demand.
• Watch for evidence of on-device AI adoption and monetization in the next 6 months; if Apple demonstrates strong uptake without margin dilution, trim any relative underweight in AAPL and rotate out of overowned AI infrastructure exposures.