Lamb Labs (YC S26)

20,000tok/s, custom silicon chips, building the future of hardware for ai inference! @ycombinator S26 🐑 🐑 @KotechaNiki @thomasslanning

[Lamb Labs (YC S26) @LambLabs]

Lamb Labs (YC S26) is building custom AI inference chips that can do 20,000+ tok/s at 63x higher Intelligence per Watt (IPW) than traditional GPUs.

GPUs became the default for AI because they were available, not because they were the right hardware.

[Lamb Labs (YC S26) @LambLabs]

@amritessh re: hardcoding model weights, our bet is that models will start stabilizing plus for a lot of tasks, a lot of the models already do a good job, i.e., you don’t always need the frontier/SOTA model…but also automating the chip design pipeline!

[Lamb Labs (YC S26) @LambLabs]

Lamb Labs (YC S26) is building custom AI inference chips that can do 20,000+ tok/s at 63x higher Intelligence per Watt (IPW) than traditional GPUs.

GPUs became the default for AI because they were available, not because they were the right hardware.

[Lamb Labs (YC S26) @LambLabs]

@amritessh We will be launching this next week, our part 2 software launch, this will include an open source model too! 🐑

[Lamb Labs (YC S26) @LambLabs]

@mj_hiremath @ycombinator creds to @ElevenLabs 🐑

[Lamb Labs (YC S26) @LambLabs]

@OpinionAILtd @KotechaNiki @wazidotapp @sama

[ZAZA @OpinionAILtd]

@KotechaNiki @wazidotapp Singularity is here with @LambLabs

[Lamb Labs (YC S26) @LambLabs]

@ouranoscapital A lot 🫣, but lots of ways of reducing cost

[Lamb Labs (YC S26) @LambLabs]

@mohitwt_ All of them

[Berat Celik @beratcelik0]

A $100M GPU deal doesn't close on an exchange. It closes in a group chat, off a voice note.

Today we're launching Stoa: an RFQ marketplace for GPUs. $300M+ in RFQs in our first month.

Post what you need. Vetted dealers bid blind. Firm quotes within 48 hours. We handle KYB, contracts, shipping, and settlement.

Every RFQ, firm quote, award, and settlement runs through Stoa, so we see what hardware trades for. That history becomes prices buyers, sellers, and lenders can mark, finance, and hedge against.

@stoaexchange

[Lamb Labs (YC S26) @LambLabs]

busy lambs @ycombinator

[Niki Kotecha @KotechaNiki]

spot the lambs outside of the HQ, on a road trip to meet customers 🐑🐑

[Niki Kotecha @KotechaNiki]

spot the lambs outside of the HQ, on a road trip to meet customers 🐑🐑

[tyler hogge @thogge]

OpenAI has signed now.

What a fascinating day of game theory.

[Lamb Labs (YC S26) @LambLabs]

talking about chips, silicon and physical ai at the Advancing in AI AMD conference!

[Lamb Labs (YC S26) @LambLabs]

Getting the board warmed up! 6W for a 7B param model! We’re only going to get better from here

[Lamb Labs (YC S26) @LambLabs]

Hardware is tough, our dev board finally showed up! Time to cook over the weekend. 🤖🦾 🐑🐑

[Arena.ai @arena]

Big news: Kimi-K3 by @Kimi_Moonshot is now #1 in the Frontend Code Arena with 1679 pts, surpassing Claude Fable 5.

This is a 17-place jump from Kimi-k2.6 (#18 -> #1).

In Frontend, Kimi-K3 ranked #1 in 6 of 7 domains: Brand & Marketing, Reference-Based Design, Data & Analytics, Consumer Product, Simulations, and Content Creation Tools, landing #2 only in Gaming behind Fable 5.

The full model weights will be released by July 27.

Congrats to the @Kimi_Moonshot team on this major milestone!

[Lamb Labs (YC S26) @LambLabs]

If you’re deploying AI at the edge or thinking about custom hardware for your models, drop us a message, we’d love to chat!

[Lamb Labs (YC S26) @LambLabs]

The team attended the #AUTONOMOUS2026 conference and had some great conversations around intelligence per watt, Physical AI, and bringing intelligence to the edge.

As Physical AI moves from demos to real world deployment, the hardware it’s running on matters.

[Lamb Labs (YC S26) @LambLabs]

world cup vs 🐑 founders

[Lamb Labs (YC S26) @LambLabs]

@yousefhll @nourzahzah @runvendo @ycombinator lets go team!!!