1. Home
  2. Companies
  3. Base Compute

About

Base Compute is an AI inference lab that builds inference engines and infrastructure designed to run AGI on-device. Its work centres on making powerful AI operate locally on users' own hardware, so that data never leaves the device and inference can be delivered at near-zero marginal cost. The company's technical domains include AI inference engine development, AI infrastructure, systems optimisation, on-device AI and automated AI research.

The team is small and senior, composed of researchers and engineers. It works close to the metal and from first principles, with an emphasis on deep systems work and on building foundational systems that ship. Research receives substantial investment, particularly in automated AI research.

Base Compute is headquartered in Melbourne, Australia, and also operates from Berlin, Germany. Further information is available at https://basecompute.co/.

Similar companies

Fractile logoFR

Fractile

Fractile is a UK-based semiconductor company building AI acceleration hardware to radically improve frontier model inference performance.

Eridu AI logoEA

Eridu AI

Silicon Valley hardware startup building infrastructure solutions to accelerate AI model training and inference in data centres.

webAI logoWE

webAI

webAI provides a private AI platform that enables enterprises and governments to build and operate custom AI on their own local infrastructure with full data ownership.

Gimlet Labs logoGL

Gimlet Labs

Gimlet Labs is an applied research lab building AI infrastructure - including serverless inference and autonomous kernel generation - to make AI workloads 10X more efficient at datacenter scale.

Inference logoIN

Inference

Inference.net runs a distributed GPU cluster providing AI inference infrastructure, custom model training, and deployment services designed to reduce costs and improve speed for production AI workloads.

d-Matrix logoD-

d-Matrix

d-Matrix designs purpose-built AI inference computing platforms, using digital in-memory compute technology to run generative AI at scale efficiently and cost-effectively.