Cerebras IPO Ignites AI Chip Wave|Applied Materials Signals How Long It Lasts

AI Chip IPO Surge

Cerebras priced its IPO at $185 a share on Wednesday night and opened Thursday at $350, closing up 68% — but that gap between pricing and opening tells a more specific story than the headline percentage. The price was set above an already elevated range, underwriters still couldn't satisfy demand, and the stock opened 89% above that elevated price before settling. That sequence means institutional allocators — not just retail — were chasing exposure on day one, which is a different signal than a retail-driven pop.

The company's business model explains why. Cerebras builds Wafer Scale Engine chips using the entire silicon wafer rather than the small dies stitched together in Nvidia's GPUs. That architecture gives its inference hardware a claimed 15x speed advantage for the specific task of running AI models in production — not training them. AI training spending is well-covered by Nvidia. AI inference, the process of actually serving model outputs to billions of users, is where the next capacity build-out is concentrating. OpenAI signed a $20 billion multi-year deal with Cerebras in January precisely because it has more user demand than compute supply.

The capital flow on day one was institutional and directional: Benchmark's stake reached $5.5 billion at close, Foundation Capital's $4.8 billion — venture investors who entered in 2016 saw a liquidity event that returns multiples of their entire fund. That exit pressure is also a forward variable. Lock-up expiration in several months will push meaningful share supply back into the market at a 187x trailing revenue multiple. What the IPO proved is that the AI infrastructure conviction trade is intact. What it did not prove is that Cerebras is exempt from the same multiple compression that follows any supply unlock into a concentrated position.

Equipment Cycle Depth

The Cerebras IPO confirmed investor appetite for the end product of AI compute. Applied Materials reported the results that explain how that appetite is being manufactured. The semiconductor equipment company posted Q2 revenue of $7.91 billion, 11.4% above the prior year and 2.7% above consensus, and guided Q3 to $8.95 billion — a number 9.2% above what analysts were modeling. CEO Gary Dickerson said the semiconductor equipment business is now expected to grow more than 30% in calendar 2026.

That guidance revision is the signal that matters. Applied Materials doesn't sell chips — it sells the machines that make chips. When it raises forward guidance by 9% above a 22% consensus growth forecast, it means its customers — TSMC, Samsung, Intel, memory fabs — are placing orders that extend beyond the current quarter's demand. Lynx Equity shifted its valuation focus to 2028 earnings, projecting 35% systems revenue growth in 2027 and a new NAND investment cycle in 2028. Citi's analyst linked the guidance directly to hyperscaler capex: data center operators are pulling forward semiconductor manufacturing equipment orders because the inference capacity Cerebras is being paid to deliver requires the fabs to run at full capacity now.

The capital flow here runs from institutional investors into a stock that had already gained 70% year to date and was testing 52-week highs going into earnings — not a classic rotation trade, but a momentum confirmation from a company whose order book extends visibility into 2028. The counter-signal is the gross margin trajectory: free cash flow margin compressed to 2.7% from 14.9% a year ago, meaning the revenue growth is consuming working capital at an accelerating rate. Applied Materials is building capacity to serve the AI cycle, but the cost of building that capacity is running ahead of the cash it generates — a condition that narrows if pricing holds, and widens if customers slow orders.

The Retail-Led Rally's Limit

The Cerebras IPO and the Applied Materials guidance revision did not happen in a vacuum. They happened on the same day the S&P 500 touched 7,500 and the Dow crossed 50,000 — milestones that were not driven by institutional rotation but by retail re-entry after a pullback caused by the U.S.-Iran war earlier this year. Goldman Sachs found that retail trading volumes surged 28% since mid-April, with Mom-and-Pop investors returning to their high-growth mainstays after sitting out March and parts of April.

That retail return amplifies the move but also concentrates the fragility. More than 70% of the 2026 earnings estimate revisions driving the S&P 500 are coming from just six companies. Retail investors hold roughly 10% of total equity market value but account for 20% of daily trading volume because they use margin and leveraged ETFs at higher rates than institutional players — meaning their flows punch above their capital weight. Goldman's options desk flagged that the correlation between the Nasdaq 100 index and its one-month call option price has gone positive for only the fourth time in a decade, a dynamic Goldman labeled an "up crash." The prior three instances produced an average 2.7% additional return over the following month — but the last comparable episode in early 2018 was followed by Volmageddon, when the VIX spiked to 50 and short-volatility products imploded.

The conditions for that reversal were a Fed policy shift that markets had not yet priced. Today's equivalent is already partially on the table: money markets have fully priced out any Fed rate cut in 2026 and are assigning a 28% probability to a December hike, after April CPI came in at 3.8% and PPI posted its hottest print since 2022. Kevin Warsh was confirmed as Fed chair on Wednesday — the first chair transition since 2018 — and his institutional reputation leans toward tighter policy than the prior regime. The benchmark to watch is the 10-year Treasury yield, which touched its highest level since July at 4.45% before easing slightly on Thursday. If it reclaims and holds above 4.5%, the multiple that retail investors paid for Cerebras at 187x trailing revenue and for the S&P 500 at a 40% premium to global peers becomes harder to sustain. The rally's internal logic depends on AI earnings growth outrunning the rising discount rate — and Applied Materials' free cash flow compression is one early signal that the cost side of that equation is not yet resolved.