Fed Hike Odds Surge|Can the AI IPO Wave Hold?

Bond Market Warning

The 30-year Treasury yield closed at 5.13% on Friday — its highest level since June 2007. That threshold matters not because of the round number, but because the last two times long bonds crossed 5%, the Fed's ability to hold its policy stance became the dominant market variable, not earnings, not growth.

What forced yields there this week was a two-punch inflation release. April CPI came in at 3.8% year over year, up from 3.3% in March — the highest reading since May 2023. Then the Producer Price Index printed at 6% annually, its steepest climb since December 2023. Consumer spending held firm, with retail sales up 0.5% month over month in April. That combination — rising prices and no demand slowdown — removes the one argument the Fed had for holding rates steady.

The market's position had been clear all year. Earlier in 2026, futures traders priced three rate cuts. That fell to two, then one, then the debate became whether the Fed simply holds. As of Friday's close, the CME FedWatch Tool put the probability of a June cut at below 1%. By December, traders are now pricing nearly a 50% chance of a rate hike — not a hold. The direction has flipped entirely.

The mechanism connecting Trump's Beijing summit to that yield move is the piece most headlines missed. US officials entered the China trip hoping Xi would pressure Iran to reopen the Strait of Hormuz. Trump left without a concrete agreement. Oil prices rose on Friday afternoon as that outcome became clear, reinforcing energy-driven inflation as a persistent input rather than a temporary shock. Retail bond sellers and institutional duration managers absorbed the implications simultaneously — Treasury prices fell across the curve as sellers overwhelmed buyers, with the 10-year yield climbing 13 basis points to 4.59%, also its highest since May 2025.

Kevin Warsh was confirmed as the new Fed chair the same morning the data landed. Markets did not interpret the leadership transition as a policy reprieve. Futures pricing after the confirmation showed no softening in rate hike odds — suggesting traders read Warsh's ascent as accelerating the shift away from accommodation, not slowing it.

The bond market is now pricing a scenario the equity market has not fully absorbed: that the next Fed move is up, not down, and that the timeline is measured in months, not years.

Metals Capital Exit

If the bond market's message is that the Fed's next move is a hike, then the move in precious metals on Friday is its confirmation. Gold fell 2.25% to approximately $4,545, its lowest close in over a week. Silver dropped nearly 8%, erasing all gains from earlier in the week. Both moves happened in a single session — not gradual repositioning, but rapid exit.

The mechanism is straightforward but the magnitude is the signal. Non-yielding assets like gold and silver carry an embedded opportunity cost that rises with interest rates. When the December hike probability moved from 33% to 45% in 24 hours — as it did between Thursday and Friday this week — that cost change reprices immediately in metals, even before the Fed acts. Traders are not waiting for an actual hike; they are repositioning for the probability of one.

The dollar's move compounded the exit. The US Dollar Index crossed above 99.00 on Friday, its highest level since April 8. A stronger dollar creates a second headwind for metals priced in dollars — it raises the effective cost for non-US buyers, suppressing foreign demand at the same moment domestic buyers are already reducing exposure. The combination produced a synchronized selloff, not a rotation: capital moved out of both gold and silver simultaneously into dollar-denominated cash and short-term Treasuries, as interpreted from volume and price action across both metals on Friday.

Gold's technical position adds context the price alone doesn't convey. XAU/USD closed below its 20-day moving average on Friday, with the RSI at 40 — not yet oversold, meaning sellers still have room before a technical bounce forces the other direction. Silver held above its 200-day moving average at $65, but that lower band is far enough away that bears retain control without a near-term catalyst reversal.

The forward conditional is narrow: if the December hike probability holds above 45% and oil prices do not retreat meaningfully — which requires either a Strait of Hormuz resolution or a demand destruction signal — gold's next structural support sits near $4,350, and silver faces the gap between current levels near $76 and the 100-day moving average at $81, now acting as overhead resistance rather than support.

The metal move on Friday is not a safe-haven failure — it is the market pricing a new regime where the Fed's next action raises the floor on the risk-free rate. The question that settles whether this is a temporary dislocation or a structural rerating is whether the inflation data that forced yield repricing this week continues to confirm in June.

AI IPO Demand Test

The same macro session that hammered gold and sent Treasury yields to near-20-year highs also produced the largest IPO of 2026 — and it priced at $185 per share and closed its first day at $311. Cerebras Systems, an AI inference chip company, surged 68% in its Nasdaq debut on Thursday, with demand exceeding available shares by more than 20 times. On Friday, while the bond market repriced the entire rate outlook, Cerebras shares fell roughly 10% — and yet still held above $280. That combination tells a more specific story than the headline suggests.

The Cerebras debut is a capital flow event with a precise transmission. For most of the AI rally, the relevant trade was concentrated in a small group: Nvidia on the hardware side, Microsoft and Alphabet on the platform side. The Cerebras IPO represents the first successful redirection of that AI capital into pure-play inference infrastructure outside the Magnificent Seven. Investors who were priced out of Nvidia at current valuations, or who wanted more direct exposure to the inference layer rather than to hyperscaler margins, found a new entry point. The 20x oversubscription rate was not retail enthusiasm — institutional demand at that scale requires deliberate allocation decisions by funds running hundreds of millions.

The counter-signal on Friday — the 10% decline after Thursday's surge — does not erase the signal from the first day. It is consistent with the macro headwind: a 50% implied probability of a December rate hike raises the discount rate applied to high-growth, pre-profit or early-profit technology companies. Cerebras is sold out of manufacturing capacity into 2027 and carries a $37 billion annualized AI cloud revenue run rate at OpenAI alone under its contract terms — but at $280 per share, it is trading at roughly 4x the valuation it commanded at IPO pricing in less than 48 hours.

The SpaceX and OpenAI IPO pipeline now faces a direct read-through from the Cerebras experience. Cerebras priced at $185 with a $5.5 billion raise — exceeding ARM Holdings' prior record. SpaceX is targeting a June 11 Nasdaq pricing at a $1.75 trillion valuation. If Cerebras' Friday decline was driven by macro repricing rather than company-specific doubt, the AI IPO window remains open but has a new ceiling: investors are willing to pay for pure-play AI infrastructure, but only if the inflation and rate trajectory does not worsen from here.

The verification condition is specific: if NVIDIA's May 20 earnings report — which Polymarket traders assign 95% odds of beating — confirms inference chip demand is expanding at the rate Cerebras' CFO described when saying the company is "sold out into 2027," the Cerebras decline on Friday reads as a macro dip rather than a demand rejection. If NVIDIA's data center revenue or forward guidance disappoints, the Cerebras valuation reset on Friday was only the beginning, and the SpaceX IPO pricing conversation shifts accordingly.