The chip that ARM wants to buy doubled today.

$Cerebras(CBRS.US) Can you imagine a chip as big as an iPad?

A normal GPU, like the NVIDIA H100, is about the size of a postage stamp. But there's a company dedicated to "not cutting the wafer"—Cerebras. They make an entire 12-inch wafer (about 30 cm in diameter) into a single chip, called the Wafer-Scale Engine (WSE). It packs 900,000 cores and can run AI inference at 10-20 times the speed of a GPU.

This company just had its IPO on Nasdaq, ticker CBRS. IPO priced at 5, opened at 5, closed at 1, up +68.14% for the day. It even surged to around 9 after hours. This is Nasdaq's largest IPO in 2026, raising $5.55 billion. Cathie Wood's ARK bought in immediately.

Why is the market so crazy? Because Cerebras is currently the only AI chip company directly challenging NVIDIA and whose products are already on sale. Groq is doing it, SambaNova is doing it, ASIC vendors are competing, but Cerebras is the first to actually have major customers, actual revenue, and can truly demonstrate solutions that are many times faster than H100 on-site.

ARM and SoftBank even tried to buy the whole company a few weeks before the IPO and were rejected—you can imagine how sought-after this company is in AI circles.

But I have to pour some cold water on this:

First, over 80%+ of this stock's revenue comes from a single UAE client called G42. This is like a listed company whose main revenue depends on a friend sending you money every month. If that friend stops cooperating one day, the company gets cut in half.

Second, the valuation has already priced in 5 years of imagination. P/E is 761x, and the price-to-book ratio is negative (the company isn't truly profitable yet), with a market cap of $0 billion. An IPO at this price level is very similar to Snowflake's listing in 2020—a great story, but the price is already overextended.

Third, for this kind of single-track AI star IPO stock, a 40-50% pullback within six months is highly likely. From Palantir to Snowflake to ARM, there are almost no exceptions.

My judgment is: This company is worth tracking long-term, but not worth chasing on its IPO debut. If you're really bullish on the "GPU replacement" narrative, waiting for the wave of employee selling after the 6-month lock-up period expires and market sentiment cools down before entering will give you much better odds. At this current level, it's better for watching than for buying.

If you already bought in at 1 pre-market, set your stop-loss at the 5 whole number level (near the IPO price). If it breaks below that, it means the market isn't buying the IPO story anymore. Don't fantasize about "holding for the long term"—holding for the long term after a stock doubles on its IPO day is gambling, not investing.