◆ Sitemap

Every page, one place.

A directory of every page on the OXMIQ Labs site. The machine-readable version lives at sitemap.xml.

Company
Home
OXMIQ Labs — licensable GPU IP and AI runtime software.
About
Founded in 2023 by Raja Koduri. Re-architecting the GPU stack from atoms to agents.
Team
Semiconductor and AI veterans led by Raja Koduri.
Contact
Partnerships, licensing, careers, press. HQ Campbell, CA.
FAQ
Common questions on CUDA compatibility, supported hardware, licensing, and the OxCore, OxQuilt, OxPython, and OxCapsule products.
Products
OxCore
Licensable GPU IP that fuses CPU, GPU, and TPU into one unified core.
OxQuilt
Supply-chain-adaptive chiplet packaging IP. Tuned per workload, mobile to AI pods.
OxCapsule
The developer tool for running AI models on any accelerator.
OxPython
A Python-like runtime for heterogeneous silicon.
Beta
Apply for OxCapsule beta — early access to local AI inference on any accelerator.
Blog
Blog Index
News, research, and conversations on re-architecting the GPU stack.
Raja Koduri, Alchedek interview
Energy as the binding constraint, sovereign infrastructure, and the path to outcomes-as-a-service.
OXMIQ-AU Centre of Excellence in AI Systems
OXMIQ Labs and Andhra University sign an MOU to train the next generation of GPU and AI silicon talent.
Eight months out of stealth
Where each product stands eight months after launch, plus the AM Intelligence partnership and team growth.
AM Intelligence — 2 GW renewable AI compute
OXMIQ signs AM Intelligence Labs to architect 2 GW of renewable-powered AI compute by 2030.
Raja Koduri launches OXMIQ India
A new model for sovereign AI compute and India's role in AI and semiconductors.
Re-architecting the GPU stack (podcast)
OXMIQ CEO Raja Koduri on re-architecting the GPU stack — from atoms to agents.
Re-architecting the GPU stack (thesis)
Why the current GPU paradigm is breaking under AI inference economics.
OxCapsule beta launch
Run AI on any accelerator without rewrites — the developer tool from OXMIQ Labs.
Chiplet quilting
How OxQuilt makes chiplet packaging supply-chain-adaptive.
Memory economics
Why bandwidth, capacity, and access patterns dominate AI inference cost at scale.
Four RTX 6000 Blackwell GPUs
384 GB of VRAM in one machine for local LLM inference.
Legal
Privacy Policy
How we collect, use, and protect personal data.