Matt Frankel

Yeah, as I mentioned, the growth in RPOs and the AI-specific revenue growth, they're certainly encouraging.

And revenue is expected to slow over the next quarter.

But remember that that seats and credits model is going to take several quarters at a minimum to roll out to all of its customers.

So keep that in mind.

But they're not out of the woods.

So what they needed to do, they did.

They reframed the narrative from Monday is losing to AI to maybe AI could be a tailwind for Monday.

They needed to reframe the narrative like that, and they did that with this quarterly report.

But at the same time, I would place a big emphasis on that word maybe in that sentence.

I personally am not a buyer here.

It's on my radar, especially with the massive buybacks they're doing, but they are not out of the woods yet.

But solid quarter, credit where it's due.

Yeah, it's pretty clear, actually.

The bull case is the truly unique nature of Cerebus' hardware.

I mean, independent testing shows that one of their products, the CS3, delivers 21 times faster inference than NVIDIA's flagship GPU and with 32% lower costs when you include things like energy and hardware.

There's a case to be made

that Cerberus is well positioned for the shift away from training AI models, which is what we've been doing so far and why NVIDIA's GPUs have been so successful, to running AI models where speed becomes the most important thing and that's where they excel.

And that's especially true as the agentic AI boom unfolds.

I mean, the big agreements with OpenAI and AWS, Rachel mentioned, they certainly help the bull case as well.

At IPO, Cerebras is going to trade for roughly 100 times sales, assuming a $50 billion market cap, but it grew revenue by 76% year over year with a 47% gap net margin last year.