Greg Brockman

This is awesome.

This is one blade of the Frontier supercomputer.

It's the world's fastest supercomputer for open science.

This is one of somewhere over 18,000 of these in this room, all networked together to make science calculations.

Huge.

This is probably 50 times your laptop, just this one.

When we're using our laptop, we're asking it to do a lot of things.

When scientists are doing this, they are writing algorithms that have

billions of parameters.

We're working with some folks from GE designing a new aircraft engine.

The amount of parameters that they're looking at is unbathable.

And so they need those really, really fast calculations to make their simulation usable to make decisions for their business.

So when we buy these systems, the technology doesn't exist.

We make the procurement and then we design them hand in hand with the vendor.

In this case, it was Hewlett Packard Enterprise and AMD, for example.

The cooling technology here, we pushed them to make warm water cooling, saving us at least 40% in our energy costs.

And then we pushed them to design chips so that we could make them available to our scientists.

A year later, you could buy this chip.

We could run protected health data, which is a huge area for us.

We have a lot of users who come to us with huge, huge data sets, and they want to look for trends, cancer-related trends, for example, so that clinicians can make decisions way faster.