Up 108% on the first day! The biggest AI dark horse of 2026 has arrived, and Ultraman “gets rich while doing nothing” again

Author: Xinzhiyuan

This is insane! A chip the size of a dinner plate directly sparks the biggest tech IPO of 2026. On day one, it surged 108%. Not only did it make Altman nearly 10 times his money, it also fired the first shot of the ASI era’s more-than-3-trillion-IPO tsunami.

In 2026, the largest U.S. tech IPO was born!

Today, AI chip company Cerebras officially listed on NASDAQ, with a 108% surge on its first day.

The IPO priced at $185 per share, opened immediately at $385, and ended the day at $311, with a valuation that at one point reached $100 billion.

Cerebras sold 30 million shares in one go, raising $5.55 billion.

This is one of the largest IPOs for a U.S. tech company since Uber went public in 2019, with momentum even stronger than Snowflake had back then.

In the NASDAQ trading hall, Cerebras’ founder and CEO Andrew Feldman rang the opening bell.

Now, a large number of VC investors have become overnight millionaires. Altman holds 89,373 shares and turned tenfold on the spot, soaring to about $30 million.

Cerebras kicked off with a stunning start, successfully igniting the first AI IPO debut of 2026. Next, OpenAI, SpaceX, and Anthropic are lining up to enter.

The big breakout of AI IPOs is about to begin.

The biggest AI IPO of 2026: up 108% on day one

It’s hard to believe that a year ago, nobody thought this day would come.

In 2015, semiconductor industry veteran Andrew Feldman, together with experienced chip industry experts, founded Cerebras.

When Feldman began selling chips in 2019, he said something himself: “No one cares; the market isn’t ready yet.”

But if you say the path to Cerebras’ IPO is a suspense drama, that would be putting it mildly.

In September 2024, Cerebras submitted its IPO application for the first time.

As a result, it was flagged by the U.S. Committee on Foreign Investment (CFIUS)—

Because its major customer and investor is G42 from Abu Dhabi, and at that time G42 accounted for 87% of Cerebras’ revenue.

National security reviews kept getting dragged on and on. Investor confidence collapsed, and the IPO plan was forced to be shelved.

One year later, today’s story has completely reversed.

AI models finally became “smart and useful,” and Cerebras’ business directly exploded:

In 2025, revenue reached $510 million, up 76% year over year. More importantly, it turned around from a loss of $482 million to a profit of $238 million.

In just one year, from massive losses to huge gains—this turnaround left investors unable to sit still.

The customer lineup also shifted from being dominated by G42 to multiple parallel clients including OpenAI, AWS, G42, and MBZUAI.

Even more importantly, two heavyweight new customers appeared—

OpenAI signed a multi-year contract worth more than $20 billion, locking in 750 MW of Cerebras computing power, to be deployed in phases starting in 2026.

AWS announced that it would deploy Cerebras CS-3 chips in its data centers and open them to developers via Amazon Bedrock.

In a recent interview, Feldman revealed that during roadshows, he had three core points he most needed investors to believe:

First, inference demand will surge by 1 million times;

Second, getting computing power isn’t only something you can do with GPUs;

Third, the moat of CUDA is actually being exaggerated.

This time, Cerebras successfully fired the opening shot of the artificial intelligence IPO front, and its debut can only be described as stunning.

ASI-era overnight wealth—OpenAI also becomes a “big winner”

Cerebras’ IPO created a textbook-level VC exit feast.

From the seed round to ringing the bell, these institutions waited for a full ten years.

Foundation Capital: $37 million → $2.8 billion, 76x return

Foundation Capital invested about $37 million, with a 7% stake. At the IPO price of $185, it was worth $2.8 billion—76x returns.

At the first-day closing price of $311, that’s about $4.8 billion.

Benchmark: $268 million → $3.3 billion, 12x return

Benchmark also entered in 2016 during Series A, led by general partner Eric Vishria.

It invested a total of $268 million, holding 8.1%. At the IPO price, that corresponds to $3.2 billion—12x returns. At the closing price, it soared to $5.5 billion.

Eclipse Capital: $146.5 million → $2.5 billion, 17x return

Eclipse Ventures invested $146.5 million, holding 6.2%. At the IPO price, it corresponds to $2.5 billion—17x.

Eclipse’s story is even more legendary—the person who made the investment decision was Pierre Lamond, a 95-year-old veteran of “Fairchild Semiconductor.”

When at Khosla Ventures, he had already invested in Feldman’s former company SeaMicro, and later joined Eclipse.

In 2016, one year after Cerebras was founded, he decided to invest again. It was as if the Silicon Valley chip patriarch personally stamped his approval.

In 2017, he said, “Feldman is probably one of the few entrepreneurs I would invest in twice.”

Altman’s 10x “lay back and earn”—investing for just a cent

If you’re talking about the most outrageous VC returns, it’s OpenAI!

Let’s start with Altman himself.

According to court documents revealed, Altman personally invested in Cerebras as early as February 2017, when ChatGPT hadn’t even appeared yet.

As of the end of 2025, he held 89,373 shares, worth about $3.2 million at the time.

On the day of listing, the value of these shares jumped to about $30 million.

Quietly, that money multiplied by 10.

Previously, OpenAI had once considered directly acquiring Cerebras.

Although the acquisition talks didn’t materialize, the two sides reached a “mysterious agreement” on Christmas Eve—

In the coming years, if OpenAI purchased inference computing power resources worth billions of dollars from Cerebras, it could obtain up to 11% equity.

And the price OpenAI paid for these shares was less than 1 cent per share.

With the IPO price set at $185, OpenAI’s stake value instantly ballooned. Current estimates suggest an unrealized gain of about $1.8 billion.

It’s also worth noting that OpenAI co-founder Greg Brockman also personally held about 77k shares of Cerebras.

At the closing price of $311, Brockman’s personal stake was worth about $24 million.

The technical trump card: a chip the size of a dinner plate

As everyone knows, Cerebras’ core weapon is WSE-3 (Wafer Scale Engine 3).

A single chip covers an entire 300mm wafer, with an area of 46,225 square millimeters—about the size of a dinner plate.

Ordinary chips are around 800 square millimeters; WSE-3 is nearly 60 times larger.

In terms of parameters, WSE-3 delivers overwhelming dominance—

40 trillion transistors (19 times Nvidia’s B200)

900k AI-optimized cores

125 petaflops of AI computing power (28 times B200)

5nm process, manufactured by TSMC

While others solve computing power issues with “multi-card parallelism,” Cerebras directly turns an entire wafer into a giant processor.

Behind this “brutal beauty” is a key judgment: inference is the future battlefield of AI.

Training a large model might be a one-time thing.

But inference—making the model answer every question, execute every agent task—is a continuous, nonstop consumption of computing power.

Cerebras claims that their inference speed is 10 to 20 times faster than Nvidia GPU clusters.

AI IPO tsunami—$3 trillion ready to surge

Remember, Cerebras’ IPO is just the appetizer.

The real main event is still ahead—

SpaceX (including xAI): roadshows as early as June, valuation target $1.75 trillion, and plans to allocate 30% for retail investors—unprecedented in IPO history.

If it succeeds, it will be the largest IPO in human history.

OpenAI: aims to go public in Q4 2026, targeting a valuation of $1 trillion.

It has just completed a record $122 billion private funding round. In 2025, revenue was $13.1 billion, but losses reached $14 billion; profitability is expected only around 2030.

Anthropic: plans to go public in October. It has already raised $300 billion in funding, with a valuation of $900 billion. IPO proceeds could potentially exceed $60 billion.

Latest data shows its annualized revenue has surged from $9 billion at the end of 2025 to more than $44 billion.

Together, the three companies’ combined valuation is over $3 trillion.

Fortune wrote a harsh line: “SpaceX, OpenAI, and Anthropic may restart the IPO market—or drain it dry.”

If the three concentrate and go public within half a year, total fundraising could exceed $240 billion—more than the total U.S. IPO funding of most years in recent times combined.

The final showdown: when capital starts betting on ASI

But if you think Wall Street is only betting on a chip company’s financial statements, that would be too naive.

When you stretch the timeline, Cerebras’ story reveals a deeper logic: capital is placing bets on ASI in advance.

Back in 2017, Greg Brockman wrote in an email to the OpenAI team:

Exclusive access to Cerebras’ hardware will give OpenAI an overwhelming computational advantage over Google.

At that time, ChatGPT didn’t exist, and LLMs were still just concepts in academic papers—but OpenAI’s core team already knew one thing:

The path to AGI and ultimately ASI is ultimately a computing power arms race. Today in 2026, this judgment is being validated relentlessly.

What this IPO truly bets on is never just a company’s P/E ratio.

It bets on: the next intelligence humanity is building. And this intelligence requires a scale of computing power far beyond what anyone today can imagine.

When capital prices a chip company that’s only 10 years old at $66 billion, when OpenAI is willing to put up $20 billion just to lock in a 1/40 computing power gap, and when SpaceX, OpenAI, and Anthropic are lining up to prepare for trillion-dollar IPOs—

This is humanity laying the foundation in advance for the arrival of superintelligence.

Whoever lays the foundation first gets the ticket to the ASI era. Whoever hesitates will forever only be a spectator.

LAYOUT REFERENCE (source): total_lines=201, non_empty_lines=101, blank_lines=100