Kwasi Ankomah

And again, at a small scale, maybe not a huge problem.

At a big scale, with thousands of compute units, that does begin to add up.

So that...

That innovation is big.

And the reason that's big is it allows us to store bigger models.

So to give you an example of this, the DeepSeek models, so that would be DeepSeek R1 and DeepSeek V3, they're 670 billion parameters, right?

Absolutely huge.

Now, a lot of providers don't serve that model.

They can't physically.

And we can, basically because of the way that we've architected our chip.

And again, the folks that started the company, they had all of this in mind when they designed the chip.

So that's the big thing.

It allows us to run very large models.

Now, the second thing that that allows us to do is it allows us to run many models.

So that DDR bit allows us to store.

So if you imagine that you, you know, a GPU or alternative architecture could only store like this one model.

And in order for you to get another model, you need another unit of computing, another kind of, you know, let's call it a node.

Now, because of our kind of large DDR, it allows us to kind of store these other models so you can switch.

And this becomes super important for agentic applications because you might have an application that maybe uses the GPT-OSS model that we're running at the moment, or it might use a Lama 8B, but we can have those on the same node.

So your inference and hardware cost stays flat because we can go and get the model.