We've partnered with Modular to create Large Scale Inference (LSI), a new OpenAI-compatible inference service. It's up to 85% cheaper than other offerings & can handle trillion-token scale. We originally created it at the request of a major AI lab to do large scale multimodal

We're soon launching LSI, a product built with love between San Francisco Compute and Modular

I'm very excited to partner with SFCompute - Evan and team are phenomenally driven and built a powerful platform for scaling GPU solutions like never before. Combined with Modular's high-performance inference solutions, they're able to deliver incredible TCO advantages! 👇

No two days are alike - new people - brilliant minds - everyday - from all over the world.

4mo ago: We bought a used forklift & strapped an Ai kit to it 4mo later: We’re moving hundreds of pallets a day in a customers warehouse Yesterday I got 3 requests totaling 100+ forklifts We have to scale right now V2 coming soon

Cross company wins introduce the best slack emojis

I was really excited to use Silicon Data but it's just the index don't make any sense and is way more different than what the actual market is. $3.35 as the H100 hourly market price is just wrong, unless someone is breaking the law... 👀

We're going to sublease our current office space in the center of Hayes Valley. It's a beautiful, two floor spot, at about 2900 sqft in total, with room for about 20 desks. If you're interested, please DM!

Finally, Sidebar. Josh Miller youtube.com/watch?v=5vexXI…

I just made a long form with Typeform and it faced technical issues and lost everything!!!!! never going back.

Keep an eye on Decart today, It has something to do with Magic 👀

Go Decart ! Everyone at SFC is excited about this launch, this is huge for the industry, this is Magic.

Is there a summer deal going on for the Nasdaq billboard in NYC or what?

If I email people at 11:30 PM in San Francisco, they read within 5-10 minuets, no one stops working here.

We're excited to combine the unbeatable engineering of Modular with the unbeatable prices of San Francisco to make the world's best priced inference service. The world gets better when costs go down.

Qwen3-Coder is now available on Cerebras, 17x faster than on GPU providers. And it's completely free. Try it out directly in your developer flow, or signup for our virtual hackathon tomorrow. It's a $5,000 prize :) Cerebras Cline

In the GPU cloud business utilization is a misleading metric, aim for revenue based metrics. Specifically Cluster MRR and ARR. You can have a lower revenue on a higher utilization.