· Venture Capital Tracker · investment-strategies · 3 min read
Agentrys’ $24.5M: Agentic Design Automation for Chipmakers
Etna Labs led Agentrys’ $19.1M seed after MediaTek’s $5.4M pre-seed — $24.5M total for an open ADA platform that lets chip teams own agent workforces across EDA tools.
Agentrys announced $24.5 million in funding on August 26, 2026: an oversubscribed $19.1M seed led by Etna Labs, on top of a $5.4M pre-seed led by MediaTek. The San Jose company is selling Agentic Design Automation (ADA) — an open platform for chip teams to own their agent workforce inside existing EDA stacks.
Unexpected truth: the differentiation claim is not “we have an LLM that writes Verilog.” It is customer-owned agents that compound on proprietary design data, versus renting the same generic capability every competitor can buy.
Key facts
| Field | Detail |
|---|---|
| Company | Agentrys (San Jose; Austin & Taiwan offices; CEO Mark Ren — ex-NVIDIA Research / IBM; ChipNeMo lineage) |
| Total | $24.5M |
| Seed | $19.1M led by Etna Labs |
| Pre-seed | $5.4M led by MediaTek |
| Date | August 26, 2026 |
| Category claim | Agentic Design Automation (ADA) |
| Use of funds | Hire; agent-native tooling; expand verification & physical-design customers |
| Proof (company) | Spec→GDS 32-bit CPU autonomous run; >90% on NVIDIA CVDP benchmark |
Who uses the product — and for what job
Users: semiconductor design teams at fabless firms, foundries, and chip startups who already own Synopsys/Cadence/Mentor-class tools and internal flows.
Job: automate multi-step verification and physical-design work, capture tribal knowledge, and improve with every completed design — without ripping out the EDA estate.
MediaTek’s Brian Hsu quote frames the need for domain-specific agent platforms semiconductor R&D can customize — a strategic buyer’s language.
Why now
- Chip design talent is scarce while AI silicon demand is not.
- General coding agents are weak on closed-loop EDA metrics; objective evaluators (sims, formal, layout rules) make self-improvement more practical than in fuzzy enterprise workflows (Etna’s RSI framing).
- DAC-season category creation (ADA) gives buyers a name for RFPs.
Why MediaTek / Etna — portfolio fit
- MediaTek Innovation Fund: First money + semiconductor distribution signal — Agentrys is not guessing at domain specificity.
- Etna Labs: Research-driven AI investor focused on foundation models applied to science/frontier domains — underwrites the recursive improvement story once a chip strategist is already in.
- Likely founder rationale: raise from a chip strategic who understands PDK/EDA pain, then a frontier AI fund that will not force a horizontal agent pivot.
No /fund/ pages for Etna Labs or MediaTek in the directory.
Competitive map
| Player | Difference |
|---|---|
| Classic EDA vendors’ AI features | Point optimizations inside vendor lock-in |
| Horizontal coding agents | Not closed-loop on chip metrics |
| Point AI-for-EDA startups | Narrow task bots vs open workforce platform |
| Internal FAANG tools | Not productized for most fabless teams |
Practical takeaway
- Founders (deep tech AI): Lead with benchmark + end-to-end artifact (GDS), not demos.
- Investors: Prefer open platforms customers can own when switching costs sit in workflow memory, not model weights alone.
- Operators: Diligence whether agents run on your PDK/tool chain — that is the enterprise close.
Sources
Editorial note: AI tools assisted with research, structure, or drafting. Venture Capital Tracker retains human editorial responsibility for factual accuracy, relevance, and source quality before publication.