Zettascale's chip. Unlike a fixed-function GPU or TPU, an XPU is highly programmable, so it can reconfigure its dataflow for each model. That is where the efficiency comes from: dense math, sparse models and the precisions they actually use, in one architecture.

Why does energy matter this much for AI?

Because the energy cost of training and running large models is becoming the constraint on data centers. Zettascale's pitch is that a chip which moves less data saves data centers a large share of their energy and cooling bill.

Skyler Chan Ashi Dissanayake

Elias Almqvist

Co-Founder & CEO at Zettascale

Energy efficient chips for AI

Zettascale

Co-Founder & CEO

Visit

About

Zettascale builds XPUs, reconfigurable chips that optimize their dataflow for each AI model instead of running a fixed function.^{[1]Launch YCExa Laboratories: 27.6x more efficient chips for AIYC, Aug 27, 2024[2]Zettascale ComputingCompany websiteCompany} Nearly all the energy in AI goes into moving data, so the architecture is built to keep it close.^{[2]Zettascale ComputingCompany websiteCompany} Elias is self-taught and has wanted to be an inventor since he was about nine.^{[3]Deeptech DecodedElias Almqvist on being self-taught and dropping out (00:23)Episode[4]Deeptech DecodedElias Almqvist on wanting to be an inventor at nine (01:10)Episode} He left a computer science and computer engineering degree at Chalmers to build the company, first named Exa Laboratories.^{[1]Launch YCExa Laboratories: 27.6x more efficient chips for AIYC, Aug 27, 2024[5]Zettascale ComputingAnnouncing our new name, Zettascale, and progressPress release, Sep 4, 2025} Its YC launch put the XPU at up to 27.6 times more efficient than an H100 in early simulation.^{[1]Launch YCExa Laboratories: 27.6x more efficient chips for AIYC, Aug 27, 2024} Zettascale (YC S24) is backed by Soma Capital and Olive Tree Capital.^{[2]Zettascale ComputingCompany websiteCompany[6]Zettascale ComputingTeamCompany}

Co-founders

Prithvi Raj

Co-founder & CTO

Episodes

Is NVIDIA’s Moat a Psyop? 28x More Efficient AI Chips | Elias Almqvist, Zetta (YC S24)
Mar 10, 20261h 10m
Watch + transcript YouTube Spotify Apple Podcasts

The reasoning layer

Decoded before it's obvious.

FAQ

Elias Almqvist is the co-founder and CEO of Zettascale, which builds energy-efficient chips for AI.^{[7]Y CombinatorZettascale Computing company profileYC[6]Zettascale ComputingTeamCompany[8]Deeptech DecodedElias Almqvist on meeting co-founder Prithvi (47:07)Episode} He is self-taught and has wanted to be an inventor since he was about nine, and left a computer science and computer engineering degree to start the company.^{[4]Deeptech DecodedElias Almqvist on wanting to be an inventor at nine (01:10)Episode[1]Launch YCExa Laboratories: 27.6x more efficient chips for AIYC, Aug 27, 2024[9]Deeptech DecodedElias Almqvist on leaving Chalmers (56:22)Episode}

Zettascale's chip.^{[2]Zettascale ComputingCompany websiteCompany} Unlike a fixed-function GPU or TPU, an XPU is highly programmable, so it can reconfigure its dataflow for each model.^{[1]Launch YCExa Laboratories: 27.6x more efficient chips for AIYC, Aug 27, 2024} That is where the efficiency comes from: dense math, sparse models and the precisions they actually use, in one architecture.^{[1]Launch YCExa Laboratories: 27.6x more efficient chips for AIYC, Aug 27, 2024}

The company's YC launch put the XPU at up to 27.6 times more efficient than an H100 GPU, based on early simulation rather than shipped silicon.^{[1]Launch YCExa Laboratories: 27.6x more efficient chips for AIYC, Aug 27, 2024[10]Deeptech DecodedElias Almqvist on the 27.6x benchmarks being old (28:18)Episode}

Because the energy cost of training and running large models is becoming the constraint on data centers.^{[2]Zettascale ComputingCompany websiteCompany} Zettascale's pitch is that a chip which moves less data saves data centers a large share of their energy and cooling bill.^{[2]Zettascale ComputingCompany websiteCompany[1]Launch YCExa Laboratories: 27.6x more efficient chips for AIYC, Aug 27, 2024}

Yes. Zettascale Computing Corporation launched as Exa Laboratories in Y Combinator's S24 batch and later changed its name.^{[5]Zettascale ComputingAnnouncing our new name, Zettascale, and progressPress release, Sep 4, 2025[1]Launch YCExa Laboratories: 27.6x more efficient chips for AIYC, Aug 27, 2024[7]Y CombinatorZettascale Computing company profileYC}