📰 Key Takeaways

New York Fed President John Williams said in a Bloomberg interview on Tuesday that it could take more than two years for AI equipment supply to catch up with demand, adding to inflation pressure. He noted that AI infrastructure buildout is driving up prices for chips and related commodities. Asked whether AI supply and demand would balance out over the next two years, Williams said, “I think it’ll take longer,” adding this is based on the “big assumption” that “demand momentum continues as expected.”

Williams also reiterated his stance that the Fed may still need one more rate hike this year to bring down inflation: “I haven’t changed my view — one more hike might still be appropriate.” As for 2027, though, he said “it depends on how things play out,” and there’s currently no reason to think further hikes will be needed by then.

On the other hand, Williams expects AI investment to stay strong even with higher rates. He believes “investors are relatively insensitive to interest rates” because the market broadly believes AI will bring massive productivity gains and economic transformation — a conviction that’s “not entirely immune to rates, but I don’t think monetary policy will have a profound effect on it.”

The piece also noted that the AI equipment shortage could affect bitcoin miners pivoting into AI business lines, as both sides compete for chips and power infrastructure. It mentioned IREN signing a $2.8 billion new AI cloud contract in July, while miners chasing AI projects already face massive financing needs amid increasingly strict lender scrutiny. See the original article for full details.

💬 JudyAI Lab Take

What stands out to us here is that the New York Fed president is directly calling out AI equipment supply-demand imbalance as something that’ll take more than two years to fix — effectively declaring that the chip and power gap isn’t a short-term blip but a structural problem.

For the AI builder community, this is a reminder that hardware costs aren’t cooling off anytime soon, and infrastructure queuing effects will keep pushing up the bar for getting chips and power. Williams noting that investors are “relatively insensitive” to rates reflects how the market is still betting on the long-term AI productivity narrative, rather than reacting to short-term swings in monetary policy. Meanwhile, cases like bitcoin miners pivoting into AI cloud (e.g., IREN’s $2.8B contract) show resource competition has spread beyond a single industry into a cross-industry scramble, with financing scrutiny tightening right along with it — a concrete signal that supply-side pressure is spreading.

When planning AI applications or picking a cloud provider, it’s worth factoring chip and power lead-time risk into your cost estimates, not just model API pricing.

📅 Source Info

🔗 Further Reading