'Magnificent 7' earnings rush reveals AI spending surge, with hyperscaler capex set to reach $725 billion in 2026

The big four hyperscalers lifted expected AI-related capex to about $725 billion, up from roughly $670 billion entering earnings season, signaling continued heavy investment in AI infrastructure. Meta raised 2026 capex to $125 billion-$145 billion, Alphabet raised full-year capex to $180 billion-$190 billion, and Microsoft said calendar 2026 capex should reach $190 billion; Amazon kept its 2026 plan largely unchanged near $200 billion. The updates reinforce strong demand signals for chips and data centers, with Meta shares down about 6% and Alphabet up 7% after results.

Analysis

The key market signal is not simply that AI spend is rising, but that the spending curve is becoming less elastic to near-term ROI scrutiny. That matters because it pushes the value chain from a “demand visibility” trade into a “capacity lock-in” trade: hyperscalers are effectively pre-committing to multi-quarter procurement, which should keep wafer starts, networking gear, power, and data-center buildouts tight through at least 2026. The second-order effect is that the bottleneck shifts upstream from chips alone to everything that turns chips into usable compute: substrates, HBM, advanced packaging, liquid cooling, and grid interconnect equipment.

The most interesting divergence is that higher capex is being rewarded differently by the market depending on whether investors view it as monetizable growth or margin dilution. Alphabet’s re-rating suggests the market is willing to pay up when capex is paired with accelerating cloud demand, while Meta is being punished because its spend looks more like a balance-sheet choice with a longer payback and less obvious near-term revenue translation. That split should persist over the next 1-2 quarters and likely widens if cloud growth stays above expectations and ad-driven businesses show even modest operating leverage.

The contrarian risk is that the consensus may be underestimating how much of this is supply inflation, not just demand growth. If a meaningful share of the capex step-up is due to higher component pricing, then vendors may capture more of the value than the hyperscalers, but hyperscaler ROIC math compresses if pricing power at the chip level remains elevated into 2026. The real reversal catalyst is not a macro slowdown; it is evidence that incremental tokens/queries are no longer scaling revenue fast enough to justify another 10%-15% upward revision in capex guidance. That would likely show up first in margin commentary, then in a de-rating of the highest-spending platforms.

From a trading perspective, the cleanest expression is to own the beneficiaries of the bottleneck, not the spenders themselves. The setup remains supportive for NVDA/TSM on a 3-6 month horizon, but the asymmetric opportunity may be in the less obvious supply chain names tied to power and interconnects, which should have less valuation risk than the headline AI leaders. META is the most vulnerable name if the market starts demanding proof that capex is translating into durable revenue acceleration rather than just a rising depreciation overhang.

Key Decisions for Investors

• Long NVDA into the next 1-2 quarters on any post-earnings digestion; thesis is that hyperscaler pre-commitment keeps order visibility strong, with upside if supply remains tight and pricing stays firm. Risk: multiple compression if investors start pricing in peak AI spend.
• Pair trade: long TSM / short META for the next 3-6 months. TSM benefits from the capex transmission mechanism regardless of which hyperscaler wins share; META is more exposed to rising depreciation and any skepticism around payback. Risk/reward favors the supplier over the spender.
• Add a basket long in AI infrastructure ex-chips — e.g., power, cooling, and networking names — on a 6-12 month horizon. These are the second-order winners if capex keeps accelerating and are less exposed to “AI bubble” headline risk than the megacaps.
• Avoid chasing MSFT and GOOGL after the initial read-through unless cloud growth continues to outpace capex growth for another quarter. The trade here is not directionally bearish, but the risk/reward is worse after the market has already validated the spend with a higher multiple.
• Use META downside hedges, such as 3-6 month put spreads, if the stock rallies on broader market strength. The market is signaling that it will tolerate capex only while monetization is visible; any miss in ad yield or engagement could reprice the stock quickly.