Wood You Buy The Dip?

Cerebras has had a rough week.

The AI chipmaker fell for six straight trading sessions — including a 12.7% drop on the same day it unveiled its latest technology.

Cathie Wood apparently saw a different chart.

Her Ark Invest funds bought $17.2 million worth of Cerebras, just one week after making another purchase.

Wood, of course, has made a career out of betting early — and often aggressively — on technologies she thinks could reshape entire industries.

→ And Cerebras certainly fits the description.

It makes enormous AI chips designed to compete with the GPU systems Nvidia has built an empire around.

The pitch? Faster AI.

The problem? Making money.

Cerebras is growing quickly, but its losses are growing too.

Which leaves a rather interesting $17 million question:

What does Cathie Wood see in $CBRS ( ▼ 4.07% )  ?

We took a look.⇩

SPONSOR BREAK presented by Paradigm*

Altucher: Inside This Case is Elon’s Next Major Wealth Boom

James Altucher – the man who predicted the rise of SpaceX years in advance – has just released a shocking new prediction about Elon Musk.
 
Inside, he explains why he thinks Elon’s latest project will be even BIGGER than SpaceX – and create up to 1.8 million new millionaires over the coming years starting Sept. 25th.
 
The pieces behind all of it, he says, are sitting inside this case.
 
Click here now for all the details.
 

Buying Into the $CBRS Slide↓

Other Headlines sponsored by MarketWise*

• Elon Musk, Peter Thiel, Sam Altman Back New Potential $367 Trillion “Medical AI” › Click here to see the stocks that could soar as “Medical AI” goes online nationwide.

• Is Trump Rewriting the Declaration of Independence? › Click here to see how it could affect you.

• Major Buy Alert Issued for August 31st → Here are the stocks you need to buy for the chance to profit.

• World’s Largest Investors Are Moving Their Money (Not Into AI) › Click to watch.

So, What Does Cerebras Actually Do?

Most AI systems get their computing power by connecting lots of smaller GPUs together.

Cerebras took a very different approach:

Build one enormous chip.

Its processors use nearly an entire silicon wafer, packing compute, memory and bandwidth together in an attempt to remove the bottlenecks that come from moving data between multiple GPUs.

The pitch? Speed.

According to Cerebras, its newest CS-4 system can deliver:

→ 4,400+ tokens per second, per user
→ Up to 30× faster inference than GPU-based systems

✱ Those are Cerebras’ own benchmarks, rather than independently verified results.

But there is one rather large vote of confidence:

OpenAI has agreed to add 750 megawatts of Cerebras computing capacity through 2028 — a deal Cerebras values at more than $20 billion.

For a company trying to prove there’s another way to build AI infrastructure, that’s a pretty serious customer to have on the order book.

SPONSOR BREAK presented by MarketWise*

“I called Tesla in 2019. Here’s what I’m buying now.”

Everyone said Tesla would go bankrupt. Luke Lango bought it anyway — and readers who followed saw 22X gains. Now Elon’s next move is brewing, and it’s bigger than Tesla and SpaceX combined.
Get the Name & Ticker — Free

Financial Picture—Genuinely Mixed?

✱ Fast growth…expensive growth.

On the surface, Cerebras is growing at the kind of pace you’d expect investors to love.

→ +74% › Q2 revenue, to $180.1M
→ 2×+ › Core sales, to $209.9M
→ +281% › Cloud revenue, to $126M
→ $880–890M › New 2026 forecast

And yet, the stock fell 16% after earnings.

The reason? Growing this fast is proving expensive.

Core gross margin fell from 46.5% → 40.6%, while Cerebras posted a $450.5 million net loss. Renting additional computing capacity to keep up with demand squeezed margins further.

So, Cerebras clearly can find demand.

The question is whether it can turn all that demand into a business that actually scales profitably.

And with Nvidia already showing just how profitable AI chips can become, that’s a pretty high bar to clear.

Wood’s AI Math:

Cerebras is only one piece of Cathie Wood’s much larger AI thesis.

And the thesis is surprisingly simple:

AI gets cheaper → companies use more of it.

Ark estimates AI training costs are falling nearly 75% a year, while the cost of running some models has dropped by as much as 99% annually.

If that continues, AI becomes affordable for far more businesses — and all those models still need somewhere to run.

→ ~$500B › Data-center system investment in 2025
→ ~$1.4T › Ark’s projection for 2030

Wood thinks the impact could stretch well beyond tech. She has argued that AI, combined with robotics and other emerging technologies, could eventually push U.S. productivity growth toward 4–6% a year.

So the Cerebras bet is a bet that the AI pie gets big enough for a lot more companies to eat.

Wood’s Portfolio↓

Source: Cathie’s Ark · ARKK portfolio weights as of Aug. 27, 2026

Even after buying Cerebras twice in one week, Wood has kept the position relatively small.

At 2.58%, it’s roughly one-quarter the size of Tesla and sits below other AI-related bets including Palantir, AMD and CoreWeave.

But there’s an interesting wrinkle:

✱ Cerebras is already a bigger ARK position than Nvidia.

She’s spreading her chips across the companies she thinks could benefit if the AI market becomes big enough to produce more than one winner.

Lesson Of The Day:

Was this email forwarded to you? Don’t miss out on future stories — subscribe using the button below.

Also, help your friends blossom this spring! Share us with them.

💬 We Want To Hear Your Story:

Got a market or stock you want us to analyze next?

Just drop your request in the comments here.

P.S. – If you no longer want to receive occasional emails from us and you want to unsubscribe, click here 👉 “Unsubscribe” . Thank you!