Skip to content
ElastixAI

Accelerated ML through co-optimization of hardware, ML, and software.

We’re bringing models, software, and hardware together to rethink how AI runs.

About Us

ElastixAI delivers elastic, cost-efficient AI inference by co-designing machine learning, system software, and reconfigurable hardware as a unified architecture.

Efficiency at every layer

Serving AI at scale takes significant compute and power. We align model optimizations, software orchestration, and reconfigurable hardware to use resources where they matter. The goal: lower cost and energy per token while maintaining throughput.

Built to keep adapting

New models bring new demands. Our co-designed stack lets us reconfigure compute as those demands change, so deployments can evolve without waiting for a new generation of fixed silicon.

Meet the people building it.

Our leadership brings together experience in AI systems, hardware architecture, and infrastructure—from research to real-world deployment.

  • Portrait of Saman Naderiparizi

    Saman Naderiparizi

    Co-founder and Chief Technology Officer

    LinkedIn
  • Portrait of Mahyar Najibi

    Mahyar Najibi

    Co-Founder and Chief Scientific Officer

    LinkedIn
  • Portrait of Mohammad Rastegari

    Mohammad Rastegari

    Co-Founder and Chief Executive Officer

    LinkedIn
  • Portrait of James Allard

    James Allard

    Chief Financial Officer / Chief Operating Officer

    LinkedIn
  • Portrait of Jon Gelsey

    Jon Gelsey

    Strategy

    LinkedIn

What’s new at ElastixAI

All blog posts
An ElastixAI title card reading “From Telephones to Transformers: A Brief History of Quantization”.

From Telephones to Transformers: A Brief History of Quantization

Model quality used to be AI’s limiting factor. Now, it’s model economics. Where cost per token, watts per rack, and tokens per second dominate the conversation, no topic is more pertinent than model optimizations. And, in that conversation, few optimizations are as well-known, and as misunderstood, as quantization.

A processor chip with a brain motif etched onto its surface, on a green-lit circuit board.

The AI Data Center Economy Runs on Tokens per Second per Watt

In early 2026, Microsoft disclosed an $80 billion unfulfilled Azure order backlog due to insufficient electricity to power the deployments. CEO Satya Nadella confirmed on the record that their GPUs are sitting idle in inventory, waiting for installations that can’t proceed until the grid can deliver the load.

A green mountain diagram: 100X more tokens rising above the GPU-based cost barrier and today's economic feasibility zone.

What Becomes Possible With 100x More Tokens?

What if we could generate 100x more tokens for the same budget? What new possibilities and world-changing applications would open up if the cost of AI stopped being the deciding factor?

An iceberg with a small tip above the waterline and a vast mass below it.

The Inference Economy: LLM Inference Is Everywhere, Not Just in Your Chatbot

When most people think of Large Language Models (LLMs), they think of chatbots. However, this perspective drastically understates the true scope and technical necessity of the inference market. Inference is much more than typing a prompt and receiving a response. In reality, it’s become a basic compute primitive that powers the entire AI lifecycle.

Keep up with what’s next.

Research, product updates, and a closer look at efficient AI inference.

Explore our insights