Big Tech has committed $660B in combined AI capital expenditure for 2026, pushing the PHLX Semiconductor Index to an 18-consecutive-day winning streak—its longest on record.1
Meta leads the individual announcements with a $100B+ chip supply agreement with AMD spanning five years.1 Amazon, Alphabet, and Microsoft are all active NVIDIA customers, anchoring demand across the sector.1
Bloomberg Intelligence projects 57% semiconductor revenue growth for 2026—twice the rate of the broader market.1 That gap reflects direct capex translation: hyperscaler infrastructure spending flows almost immediately into chip orders.
The capital allocation logic is straightforward. AI model training and inference require purpose-built silicon at scale. Each dollar of hyperscaler capex committed to AI infrastructure generates downstream demand for GPUs, custom ASICs, and high-bandwidth memory. With $660B locked in across the major platforms, chip suppliers face a multi-year contracted demand base, not a speculative one.
Balance sheet strategy is shifting accordingly. Big Tech companies are front-loading AI infrastructure costs to secure supply and lock in pricing, accepting near-term earnings pressure in exchange for long-term compute capacity. Meta's five-year AMD deal exemplifies this approach: extended supply contracts reduce procurement risk and give suppliers revenue visibility to justify their own capital expansion.
For semiconductor equities, hyperscaler capex guidance has become the primary leading indicator. Upward revisions in quarterly capex from Amazon, Alphabet, Microsoft, or Meta tend to lift the PHLX Index within days. The inverse risk is equally direct: any downward revision to AI infrastructure spending could break the current momentum fast.
The 18-day streak reflects a market pricing in sustained, not cyclical, demand. Investors are treating hyperscaler AI capex commitments as durable revenue signals for chip companies—a structural re-rating, not a momentum trade.
Whether that thesis holds depends entirely on whether the $660B in announced spending translates into actual purchase orders on schedule. So far, the contract pipeline—anchored by deals like Meta-AMD—suggests it will.1
