New York Fed President John Williams has said the supply squeeze created by the artificial intelligence buildout is likely to last beyond two years, a timeline that puts the technology cycle squarely inside the Federal Reserve’s policy horizon. Williams made the comments while reaffirming his view that the US central bank may need to raise interest rates once more this year, according to the original report. The combination matters for markets because it links the most powerful capital spending theme in equities to the cost of money.
The mechanism Williams describes is straightforward. AI data centers require enormous quantities of advanced chips, high-bandwidth memory, power generation, cooling systems and skilled construction labor. When demand for those inputs outruns the ability of suppliers to produce them, prices for the scarce components rise, and those higher costs feed through the economy rather than being absorbed quietly. A supply squeeze that runs for years, rather than quarters, keeps that pressure alive for longer than a single rate decision can address.
Why the Fed Cares About Chip Capacity
Central bankers normally look through sector-specific bottlenecks, treating them as relative price shifts that resolve on their own. The AI case is harder to dismiss. The scale of planned data center construction is large enough to move national investment, electricity demand and demand for specialized labor. If the constraint sits on the supply side, tighter monetary policy does little to relieve it directly; it works instead by cooling the broader demand that competes for the same scarce resources. That is the uncomfortable logic behind Williams’ framing: rate policy can slow the race for capacity without building any of it.
For equity investors, the read-through is mixed. Companies that own scarce capacity, whether in leading-edge logic, memory or power, retain pricing power for as long as the shortage lasts. Companies that must buy those inputs face margin pressure and may delay projects. The two-year-plus timeline Williams sketched is longer than most consensus supply forecasts, which typically assume new fabrication and power capacity arrives sooner. If he is right, the winners and losers of the AI trade stay separated for longer.
Rate Risk Meets an Already Stretched Trade
The hawkish element of the message is the more immediate market risk. A further rate increase this year would lift the discount rate applied to long-duration growth earnings, precisely the profile of the companies most exposed to AI capital spending. It would also strengthen the dollar and tighten financial conditions more broadly. Williams’ stance is not a commitment, and the Fed’s decisions depend on incoming inflation and employment data, so investors should treat it as one policymaker’s view rather than a settled path.
Cross-Asset Signals
The practical takeaway is that the AI trade and the rate cycle are now entangled. Investors watching semiconductor supply chains, power producers and data center developers should track capacity announcements alongside Fed communication, because the same shortage driving revenue growth at suppliers is also, in Williams’ view, part of what keeps inflation sticky. Until supply catches up, both sides of that equation stay in play.
Sources: cointelegraph.com · NVDA Newsroom · Reuters · Bloomberg.com · CNBC