Feds Final Stand|4 Dissents and a 120 Oil Problem

Powell's Last Call

Jerome Powell held interest rates steady for the third straight meeting on Wednesday — and four of his own colleagues voted against him. That hasn't happened since 1992. The split wasn't about whether to hold. It was about whether the Fed should even signal that cuts were coming.

Three regional presidents — Cleveland's Beth Hammack, Minneapolis' Neel Kashkari, and Dallas' Lorie Logan — voted to strip out the easing bias entirely. Governor Stephen Miran dissented the other way, pushing for a cut. The result was an 8-to-4 fracture, the most divided the Federal Open Market Committee has been in over three decades.

The reason they're split is sitting in every gas station in America. Oil has surged toward $120 a barrel as the U.S.-Iran war stretches past two months. Gas prices have climbed to $4.23 nationally. Inflation, already stuck above 3% since the end of 2023, is being pushed higher by a supply shock that shows no signs of reversing. Powell himself said at his press conference that the energy spike hasn't even peaked yet — and that monetary policy works with "long and variable lags," meaning even if the Fed wanted to act, the timing wouldn't help.

So the Fed is frozen. It cannot cut without risking inflation expectations becoming unanchored. It cannot raise without choking an economy already absorbing an oil shock. Powell described the posture as being "in a good place to move in any direction" — which is another way of saying: the data has to move first.

What makes the dissent historically significant is the context. Powell's term as chair expires May 15. Kevin Warsh, Trump's nominee to replace him, cleared the Senate Banking Committee on Wednesday. Warsh used his confirmation testimony to sharply criticize the Fed's $6.6 trillion balance sheet — down from a $8.9 trillion pandemic peak, but still far above the pre-2008 level of $900 billion. His signal: shrink it further through quantitative tightening, even if short-term rates eventually fall. That combination — lower rates but a contracting balance sheet — would push long-term yields higher, compress equity valuations, and end what one analyst called "the era of easy money that fueled the longest bull market in history."

The bond market is already pricing the shift. Iran headlines drove a bond selloff in the days ahead of Powell's final press conference. If Warsh moves aggressively on balance sheet reduction after taking over, the mortgage market — already near four-week highs — would face additional pressure. The spread between equity returns and bond yields, historically the most reliable measure of stock market risk premium, would narrow further.

Powell said he intends to remain on the Board of Governors even after surrendering the chair — staying until a DOJ investigation into Fed renovations is "well and truly over." That puts him in an unusual position: watching the institution he ran for seven years adopt a philosophy he explicitly didn't pursue. The weight of evidence suggests the Fed's next move is a cut — inflation targeting still points that direction. But the timeline just stretched. If oil stays above $100 through the summer, the easing bias the dissenting trio wanted to remove will become genuinely untenable, and a rate hike discussion will reenter the room. The one number to watch: the June CPI print. If energy costs haven't rolled over by then, the four dissents from Wednesday become five.

AI Spend Verdict

The biggest quarterly earnings day in years landed Wednesday night — and the market's response didn't match the scoreboard. Microsoft beat on revenue and Azure grew 40%, accelerating from 39% the prior quarter. Alphabet posted $109.9 billion in Q1 revenue with Google Cloud outperforming. Amazon's AWS grew 28%, its fastest pace since 2022. Three of the four Magnificent Seven companies that reported beat on every major metric. And yet futures fell.

The contradiction is structural, not cyclical. These companies are now spending over $600 billion annually on AI infrastructure — and investors are no longer willing to assume the returns will justify it. Microsoft's capital expenditure came in at $31.9 billion for the quarter, down from $37.5 billion the previous quarter. The market treated that deceleration as a green flag: spending growth is cooling while revenue growth is holding. That's the proof-of-concept investors have been waiting for. Azure's 40% growth translates to an AI revenue run rate of $37 billion annually — up 123% year over year, Microsoft said. The thesis is working, at least for cloud infrastructure.

But Meta told a different story. Reality Labs burned another $4 billion in Q1. Guidance was described as "lackluster." The stock fell after hours despite beating on earnings. And Amazon, which posted its biggest AWS jump since 2022, also saw its stock slip — because the gains in cloud were offset by questions about free cash flow. Record AI investment is denting the cash generation that investors use to value the stock. Amazon's free cash flow is under pressure even as revenue accelerates. That's the tension the market is sitting with: the numerator is growing, but so is the denominator.

The most significant structural shift didn't come from the earnings themselves. It came from the deal reshaping the AI industry's supply chain. OpenAI this week ended its exclusive relationship with Microsoft, stripped Microsoft of its right to resell OpenAI products as the sole cloud provider, and simultaneously announced a major expansion with Amazon — including a $50 billion Amazon investment in OpenAI and a $100 billion commitment to use AWS. The migration isn't complete, but the direction is clear. OpenAI is distributing its cloud dependency, and Microsoft — which holds a 20% stake in OpenAI's for-profit entity through 2030 — is watching its most valuable strategic asset become a commodity available on a competitor's platform.

The power infrastructure angle confirms the scale of what's being built. Bloom Energy reported Q1 revenue of $751 million, up 130% year over year — 42% above estimates. Hardware sales more than tripled. The company raised its 2026 revenue forecast to a range of $3.4 billion to $3.8 billion, well above the $3.2 billion Wall Street had been modeling. Its stock surged over 28%. FuelCell Energy jumped 32% in sympathy. The AI data center buildout is large enough to create an entirely new power generation sector — and Wednesday's results suggest that sector is inflecting faster than anyone expected.

The forward question is whether the AI revenue growth rate can stay above the AI spending growth rate. Microsoft showed it can, for now. But the OpenAI defection introduces a structural risk: Microsoft built its AI moat on exclusivity, and that moat just developed a gap. If Amazon and Google continue gaining OpenAI model access at scale, Azure's competitive advantage narrows to execution speed and enterprise relationships — real advantages, but not the same as sole-provider status. The verification point is Microsoft's next Azure quarter. If growth holds above 40% without exclusive OpenAI access, the moat is real. If it decelerates, the market will reprice the AI premium embedded in MSFT's valuation, which still carries the weight of an assumption that no longer holds.