The AI Chip IPO Trading Like Its Backlog Doesn't Exist

Revenue doubled and the backlog tops $25 billion. The stock fell below its IPO price anyway.

November 07, 2026

One of 2026's biggest tech IPOs has lost most of its value since its first day of trading, even as revenue doubled and its contracted backlog passed $25 billion.

Now a lockup expiration and a November 12 earnings report land within days of each other.

Beat Rollover Rush (Sponsored)

Coverage of Executive Order #14330 focuses on what changes in 2027.

What gets missed: if you're 59½ or older, you don't have to wait.

You can move part of your 401(k) or IRA into physical gold now — tax-deferred, penalty-free, with the account's tax advantages intact.

Moving early means being ahead of the rush, not behind it.

Yes, Send Me My FREE Retirement-to-Gold Guide.

© 2026 APR Insights. All rights reserved. Tax benefits depend on individual circumstances. Always consult a tax professional. Not all retirement accounts qualify.

*This is an advertisement. If you no longer wish to receive promotional messages from this advertiser, please unsubscribe here. Or write to: 351 Paseo Nuevo, Floor 2, Santa Barbara, CA, US 93101.

Six months ago, this was the hottest tech listing in years. It priced its IPO at $185, opened near $350, and handed early buyers a one-day pop you usually only read about. If you chased it on day one, you've been underwater ever since.

Since then, the business has done almost everything you'd want. Revenue more than doubled, Oracle signed on as a cloud partner, and contracted future revenue passed $25 billion.

The stock went the other way, and two events in the coming days will help decide whether that gap is an opportunity for you or a warning.

That gap is what makes Cerebras Systems (NASDAQ: CBRS), the maker of wafer-scale AI chips built for running AI models fast, one of the strangest setups you'll find in tech.

What Just Happened

The stock has fallen to less than half its $185 IPO price and roughly three-quarters below its first-day close. Three things did the damage.

First, a short seller. On October 28, Culper Research disclosed a short position and published a report claiming Cerebras wafers wear out after roughly 1.5 to 2.5 years in the field, and that the company relies on a third-party connector it doesn't own.

Law firms followed with investor investigations within days.

On November 5, CEO Andrew Feldman published a rebuttal saying the lifetime claim is wrong, that Cerebras has never shipped a system using the connector the report described, and that the company is on its third hardware generation and moving to its fourth.

You now have two sides of a technical fight that only time in the field will settle.

Second, Wall Street got more cautious. On November 2, UBS kept its buy rating but cut its target to $250 from $330, the biggest analyst revision since the IPO. Even after the cut, that target sits far above where you can buy the stock.

Third, the whole AI chip group cracked. On November 5, the S&P 500 fell 4.4% in its worst day since 2025 as Nvidia and Qualcomm dropped more than 9%.

A young, volatile AI name like this one got hit harder than the leaders, and you felt it if you owned any of them.

Musk AI Stock (Sponsored)

What Elon is doing in AI is exactly what he’s done his entire career.

He took over mobile payments with PayPal.

Electric cars with Tesla.

Commercial space launches with SpaceX.

See the pattern

Now he’s doing it with Artificial Intelligence… and James Altucher believes the shift could be significant for investors who position early.

Nothing here is assured.

According to James Altucher, there’s one company sitting at the dead center of his plan… and it's still trading for less than $100.

Which company

James compiled the details into this FREE Millionaire Maker Masterclass.

VIEW IT NOW

Whose 2022 metaverse-fueled crash set what was then the biggest single-day market-cap wipeout in history?

Login or Subscribe to participate in polls.

What the Business Actually Does

You probably picture an AI chip as one of hundreds cut from a silicon wafer. Cerebras keeps the whole wafer as one giant processor, with compute and memory on the same piece of silicon.

That cuts the data shuffling between chips, and the company says it lets its systems answer AI requests much faster than GPU clusters.

Speed is the product you're betting on. As AI agents chain together dozens of model calls to finish one task, every bit of delay adds up.

Cerebras sells that speed two ways: as hardware systems, and increasingly as a cloud service where customers rent inference by usage. The cloud side is where your growth lives, and it nearly quadrupled from a year earlier last quarter.

Five Smaller Stocks (Sponsored)

Hyperscalers are projected to spend about $800B on AI in 2026, then $1.1T in 2027, per recent reports.

Nvidia's chips can't run the AI boom by themselves.

Each one needs servers, networking, specialized power and a data center around it.

The companies filling those orders are smaller than Nvidia.

Buy ratings and big price targets now sit on 5 stocks, all flagged for Q4 2026.

Which 5 stocks are these?

*This ad is sent on behalf of Guardian Financial Publishing LLC, doing business as Stable Financial Publishing, 1013 Centre Road Suite 403-D, Wilmington, DE 19805. Nothing here is investment, financial, legal, or tax advice, and it is not a recommendation or solicitation to buy or sell any security. Investing involves substantial risk, including loss of principal; past performance (including the Nvidia figures cited) does not guarantee future results. Analyst price targets are from published third-party research as of September–October 2026, are not forecasts by Stable Financial Publishing, and may change. Spending figures are third-party projections. Nvidia is a trademark of NVIDIA Corporation. References to Nvidia, Google, Meta and OpenAI are for informational purposes only; Stable Financial Publishing is not affiliated with, endorsed by, or sponsored by any of them. Stable Financial Publishing, its owners, employees and contributors may hold positions in securities mentioned and may trade them at any time. If you no longer wish to receive these offers. Privacy: https://guardian.pub/privacy-policy/ · Terms: https://guardian.pub/terms-of-service/

Why the Market Cares

The OpenAI contract is the backbone. In December 2025, OpenAI signed a master agreement to buy 750 megawatts of Cerebras inference capacity over three to four years, extendable to five, plus an option for up to 1.25 gigawatts more by the end of 2030.

OpenAI also lent Cerebras about $1.0 billion in January 2026 to help build out that capacity.

That deal is most of the reason Cerebras reported $25.4 billion of remaining performance obligations at the end of June, and it gives you a contracted revenue base few young chip companies can match.

The customer list keeps growing. Management names AWS, AMD, and CrowdStrike as customers and partners alongside OpenAI. On October 27, Oracle said it plans to offer Cerebras inference on Oracle Cloud Infrastructure in calendar 2027.

For you, that means three of the biggest AI buyers in the world have now put Cerebras in their stack.

Management has made a big promise. On the August call, the company said it plans to more than triple revenue in 2027.

If it delivers anything close to that, a stock trading below its IPO price will look very different in hindsight. If it walks that back, you'll know quickly.

What the Financials Are Signaling

Growth is real. Second-quarter core revenue, which strips out pass-through data center revenue and warrant accounting, more than doubled to $210 million.

GAAP revenue was $180.1 million, up 74%. Cloud and other services revenue hit $126.0 million, up 281% from a year earlier. That is the kind of growth you'd expect from a company with real demand behind it.

Margins are improving, but slowly. Core gross margin reached 41%, up about 940 basis points from a year earlier, while GAAP gross margin was just 14%.

You should watch the gap between those two numbers, because that is where the cost of the OpenAI build-out shows up.

The losses look scary but are mostly paper. Cerebras reported a GAAP net loss of $450.5 million, and $386.6 million of that was stock-based compensation tied to the IPO. That dilutes you, but it isn't cash leaving the building.

The balance sheet gives you a cushion. Cerebras ended June with $8.6 billion in cash, cash equivalents, restricted cash, and short-term investments, nearly half its current market cap of about $18 billion.

Management raised full-year core revenue guidance to $880 million to $890 million and guided third-quarter core revenue to $214 million to $216 million.

The Valuation Problem You Shouldn't Ignore

On this year's numbers, you're still paying a lot. At about $18 billion, Cerebras trades near 20 times the midpoint of its 2026 core revenue guidance.

That is a premium multiple for a company that loses money on a GAAP basis and earns thin hardware margins.

Next year's promise looks very different. If revenue really more than triples in 2027, that multiple falls into the single digits. So your whole valuation case rests on one question: do you believe the 2027 ramp?

Concentration cuts both ways. Much of that $25.4 billion backlog comes from OpenAI. If OpenAI slows its build-out or renegotiates, the number you're anchoring on shrinks fast.

What Needs to Happen Next

The lockup has to pass without a flood. The IPO lockup expires around November 9, about 180 days after the May listing, which frees early holders and employees to sell.

Insiders already sold roughly $395 million of stock over the past three months, according to MarketBeat. Heavy selling around that date is likely, so you shouldn't be surprised by a rough few sessions.

The November 12 report has to answer the short seller. Cerebras reports third-quarter results after the close on November 12, with a call at 5 p.m. ET.

You want core revenue at or above the $214 million to $216 million guide, gross margins holding near 40%, and a clear reaffirmation of the plan to triple revenue in 2027.

Management has to put the reliability question to bed. Expect analysts to press on wafer lifetime, warranty costs, and how long customers keep systems in service. Hard field data would do more for your position than any blog post.

Bear Case

The short report could be partly right. Even if the two-year claim is wrong, any sign of rising warranty costs or shorter useful lives would hit margins and the value of long contracts.

Lockup supply could swamp good news. A large wave of newly sellable shares can pressure the price for weeks, regardless of what the business reports.

OpenAI dependence is a real risk. One customer drives much of the backlog. Any change in OpenAI's spending plans flows straight to Cerebras, and you'd have little warning.

Nvidia still owns the market. GPUs dominate AI inference, and Nvidia keeps pushing faster inference products of its own. Cerebras has to keep winning on speed to justify its place.

How I'd Frame a Position

Start small and let the lockup pass. Buy a third of your target size now, and keep dry powder for any selloff around the lockup date.

Add on proof, not hope. If the November 12 report beats the guide and management reaffirms the 2027 tripling plan, add your second tranche. If guidance slips or margins fall, wait for a better entry.

Keep this a satellite position. This stock has fallen roughly three-quarters from its first-day close in six months, and it can keep swinging 10% in a day. Size it so a sharp drop doesn't push you out at the bottom.

Bottom Line

Cerebras has a doubling business, a $25 billion contracted backlog, $8.6 billion of cash, and new customers like Oracle, yet you can buy it at less than half its IPO price.

The risks are real: a short report, a lockup, and heavy reliance on OpenAI. Size it as a high-risk recovery trade and let November 12 tell you whether the market has it wrong.

Setup Scorecard

Entry Zone: $70 to $90, building in three tranches around the lockup and the November 12 report

Target: $130 to $150 over 9 to 12 months, still well below UBS's reduced $250 target

Stop Loss: Reassess below $60, or if management walks back the plan to more than triple 2027 revenue

Catalyst Timeline: IPO lockup expiration around November 9; third-quarter results after the close on November 12; Oracle Cloud availability in 2027

Confidence Level: Medium-Low. The growth, backlog, and cash are real, but an unresolved short report, lockup supply, and heavy OpenAI concentration make this a speculative call.

That's our coverage for today; thanks for reading! Reply to this email with feedback or any tech stocks you want me to check out.

Best Regards,
—Noah Zelvis
Tech Stock Insider