Ex-Intel Exec Raja Koduri’s OXMIQ Lands $35M to Build ‘Arm for AI GPUs’
Key Takeaways
- OXMIQ, the AI chip startup from Raja Koduri, has raised a $35M Series A round, taking total funding to $60M.
- The company plans to license a unified AI chip architecture, challenging Nvidia’s dominance and targeting sovereign AI programs, with a commercial launch expected by early 2027.
Mentioned
Key Intelligence
Key Facts
- 1OXMIQ raised $35 million in Series A funding, bringing its total funding to $60 million.
- 2The company was founded in 2024 by Raja Koduri, former Intel graphics chief and AMD executive, and is headquartered in Campbell, California.
- 3OXMIQ plans to license a unified GPU-CPU-tensor engine IP block called OxCore, aiming to be the 'Arm for AI GPUs.'
- 4A public beta of its software stack launched in November 2025 and is now used by 20 companies and 10 universities on nearly 300 GPUs.
- 5Koduri asserts that AI compute costs must fall by 50-100x for mass adoption, and that 92% of margins are in chip design.
- 6The initial commercial focus is India and Southeast Asia, targeting semiconductor firms, neocloud providers, and sovereign AI programs.
Co-led by Samsung Catalyst Fund and Fundomo, with MediaTek, Pegatron, and Intel Capital participating
We would want to be the Arm of this next era.
In an interview with Reuters
Analysis
For founders and VCs, OXMIQ’s $35 million Series A is more than a funding event—it’s a signal that the next wave of semiconductor disruption could come from licensing IP, not manufacturing chips. Raja Koduri, the veteran behind GPUs at Intel and AMD, aims to repeat Arm’s playbook in the AI era, assembling a $60M war chest to sell chip blueprints rather than silicon. This raises a critical question for the startup ecosystem: can a fabless architecture company crack the $100B+ AI chip market?
In a notable play for the AI semiconductor market, OXMIQ, the startup founded by former Intel graphics chief and AMD executive Raja Koduri, has secured $35 million in Series A funding to commercialize a licensable AI chip architecture. The round, co-led by Samsung Catalyst Fund and Fundomo, brings OXMIQ’s total funding to $60 million and includes backing from strategic investors MediaTek, Pegatron Venture Capital, and Intel Capital, among others. This news, announced on July 1, 2026, signals a serious bet on a new model for AI chip design—one where intellectual property is licensed to semiconductor firms, cloud providers, and governments rather than being locked into a single vendor’s integrated stack. The core idea is to collapse three traditionally separate components—the graphics processor, central processor, and tensor engine—into a unified, licensable block of IP, which the company calls OxCore. Koduri frames this as the 'Arm for AI GPUs,' directly comparing OXMIQ’s licensing approach to the way Arm Holdings revolutionized the mobile processor market. Unlike Nvidia’s model, which wraps hardware, software, and systems into a tightly controlled ecosystem, OXMIQ wants to allow customers to build custom AI processors tailored to their workloads without the billion-dollar investment required for ground-up silicon development.
The round, co-led by Samsung Catalyst Fund and Fundomo, brings OXMIQ’s total funding to $60 million and includes backing from strategic investors MediaTek, Pegatron Venture Capital, and Intel Capital, among others.
What to Watch
The context is the exploding cost of AI compute, which Koduri argues must fall by 50 to 100 times to enable mass adoption, particularly in cost-sensitive markets like India. He points out that 92% of margins reside in chip design, a clear incentive for nations and companies to own that critical piece. Already, OXMIQ reports that a public beta of its software stack, launched in November 2025, is in use by 20 companies and 10 universities across nearly 300 GPUs, demonstrating early traction. The Series A funding will be deployed to finish OxCore by late 2026 or early 2027, ahead of customer deployments and commercial tape-outs. The company is headquartered in Campbell, California, but maintains significant engineering teams in Hyderabad and Bengaluru, India, underscoring its focus on Asia as the initial market. This geographic strategy aligns with the rising wave of sovereign AI programs, where governments seek self-reliance in AI infrastructure.
The implications are multifaceted. For the semiconductor industry, OXMIQ could expand the custom chip market, currently dominated by Broadcom, Marvell, and MediaTek, by offering a shortcut through IP licensing rather than full custom designs. For cloud providers and neocloud builders, it promises a middle path between buying expensive Nvidia systems and building from scratch. If OXMIQ succeeds, it could accelerate the commoditization of AI chip design, much as Arm did for CPUs, potentially reducing Nvidia’s grip on the AI hardware stack. However, the challenge is immense: competing with billions in existing R&D, building a robust software ecosystem, and convincing risk-averse chipmakers to adopt a new architecture. The company’s ability to deliver on its promises by the end of next year will be a critical test. Looking ahead, the funding round positions OXMIQ as a wildcard in the AI chip race, with a timing that could capitalize on geopolitical shifts toward tech sovereignty and the insatiable demand for more efficient AI compute.
Sources
Sources
Based on 2 source articles- Shilpa Ranipeta (in)Ex-Intel exec Raja Koduri's OXMIQ raises $35 million to build licensable AI chip architectureJul 1, 2026
- Max A. Cherney (my)Startup Oxmiq raises $35 million to build chip architecture to lower cost of AIJul 1, 2026
How we covered this story
Every story in our startup coverage is assembled from multiple primary sources, cross-referenced for factual consistency, and scored along three independent dimensions: sentiment, operational impact, and source-cluster confidence. Single-source rumors and unverifiable claims do not pass our editorial gate. When a story shows "Verified by N sources" with N≥2, the development is independently corroborated; when N=1, we mark it explicitly so readers can weigh the signal accordingly.
Impact scoring uses a 1-10 scale weighted toward regulatory, financial, and operational consequence rather than coverage volume. A topic that runs in every outlet but moves no real decisions ranks lower than a niche regulatory filing that reshapes how operators in the startup space have to behave. Read our full methodology for the scoring rubric, our glossary for term definitions, and our trends index for the longitudinal view across the beat.
| Signal on this page | What it tells you |
|---|---|
| Verified by N sources | Independent corroboration count. N≥2 is our confidence floor; N=1 is marked explicitly. |
| Impact score (1-10) | Regulatory + financial + operational weight. 8+ signals an experienced-operator action item. |
| Sentiment | Five-tier classification trained on labeled startup-specific corpora. |
| Timeline | Where applicable, the related-events sequence that contextualizes today's development. |