AI Wiki

Explore AI Wiki

Citation and evidence

ChipAgents

14 min full readUpdated 30 references

This article's verification

Report a problem with this article

More

Use this article

Raw MarkdownExplore connections

Improve this page

Suggest editRevision history

Topics

AI AgentsAI CompaniesAI Hardware

Cite this article

ChipAgents is an American startup that builds AI agents for semiconductor design and verification. The company is legally Alpha Design AI, and it markets its product and itself under the ChipAgents name.[1][2] It was founded in 2024 by William Wang, a professor of artificial intelligence at the University of California, Santa Barbara (UCSB), who is its founder and chief executive.[1][6][15] The company's agents work with the electronic design automation (EDA) tools that chip teams already use to write register-transfer-level (RTL) code, generate testbenches and verification plans, analyze simulation waveforms and trace failing tests to their root cause.[4][16] It has raised about $134 million across a pre-seed round and three Series A tranches, the last a $60 million extension led by B Capital in July 2026.[3][6][7] In October 2026 it released Build-A-Chip-Bench, a benchmark for end-to-end chip design developed with technical input from NVIDIA.[11]

Legal nameAlpha Design AI[1]
Founded2024, Santa Barbara, California[1][6]
Founder and CEOWilliam Wang[1]
HeadquartersSan Jose, California (website, October 2026); press releases from February to October 2026 give Santa Clara[2][5][11]
R&D officeGoleta, California[2]
ProductMulti-agent platform for chip design, verification and debug; Renoir domain-specific model[9][10]
FundingAbout $134 million total, as of July 2026[6]
Lead investorsScOp Venture Capital (pre-seed), Bessemer Venture Partners (Series A), Matter Venture Partners (A1), B Capital (A2)[3][4][5][6]
Websitechipagents.ai

Expanded article table

History

Founding

ChipAgents began in Wang's AI lab at UCSB with Wang and a few of his students.[24] According to a company announcement, its first employee, Mehir Arora, met Wang through an entrepreneurship competition while he was a UCSB student.[27] A ChipAgents post on the company's history traces its roots to Wang's doctoral work on recurrent neural network language models at Carnegie Mellon University and to DeepPath, a 2017 reinforcement-learning method for reasoning over knowledge graphs that he co-authored at UCSB.[24][25] Wang has said that by 2024 these techniques had matured enough to be applied to chip design and optimization.[14]

Alpha Design AI announced ChipAgents on 8 October 2024, calling it an AI agent for chip design and verification. It also announced a $3.09 million pre-seed round led by ScOp Venture Capital, with participation from Impact Assets, Amino Capital, semiconductor executives and angel investors.[3] The announcement named John Bowers, a UCSB professor and National Academy of Engineering member, as part of the leadership team. The company now lists him as a founding adviser and board member.[1][3]

In January 2025 the company said its agent passed 152 of 156 problems (97.4%) on NVIDIA's VerilogEval-v2 Verilog-generation benchmark and 155 of 156 (99.4%) on VerilogEval-Human. These results are self-reported.[18]

Growth and Series A rounds

ChipAgents announced a $21 million Series A on 21 October 2025, bringing total funding to $24 million.[4] Bessemer Venture Partners led the round,[14] with strategic backing from Micron, MediaTek and Ericsson, continued support from ScOp and Amino Capital, and angel investments from three new advisers: Wally Rhines, former CEO of Mentor Graphics; Raúl Camposano, former CTO of Synopsys; and Jack Harding, former CEO of Cadence.[4] Erez Tsur, former CEO of Cadence Israel, also joined the advisory board.[4]

On 17 February 2026 the company announced a $50 million Series A1 led by Matter Venture Partners, which the release described as a TSMC-backed hardware venture firm. Existing investors Bessemer, Micron, MediaTek and Ericsson took part, and total funding reached $74 million.[5] The same release said ChipAgents had grown from 10 to 46 employees and had moved its headquarters from Santa Barbara to Santa Clara. Wen Hsieh of Matter joined the board, and Sandeep Bharathi, president of Marvell's Data Center Group, joined the advisory board.[5]

On 29 July 2026 the company announced a further $60 million Series A2 led by new investor B Capital, with Bessemer, Micron, MediaTek, Ericsson and ScOp participating.[6][7] ChipAgents described the A2 as expanding its Series A to $134 million and said it had raised more than $134 million since its founding in 2024.[6] The three Series A tranches ($21 million, $50 million and $60 million) add up to $131 million; $134 million matches the cumulative total including the pre-seed round. Daisy Cai of B Capital joined the board.[1] Quartz, citing Reuters, reported that ChipAgents had about 64 employees and that Wang declined to say whether NVIDIA was an investor.[7]

RoundAnnouncedAmountLeadOther named participants
Pre-seed8 October 2024$3.09 millionScOp Venture CapitalImpact Assets, Amino Capital, semiconductor executives, angels[3]
Series A21 October 2025$21 millionBessemer Venture PartnersMicron, MediaTek, Ericsson, ScOp, Amino Capital; angels Rhines, Camposano, Harding[4]
Series A117 February 2026$50 millionMatter Venture PartnersBessemer, Micron, MediaTek, Ericsson[5]
Series A229 July 2026$60 millionB CapitalBessemer, Micron, MediaTek, Ericsson, ScOp[6][7]

Expanded article table

Company-reported traction

ChipAgents has published growth figures in each funding announcement. They are the company's own numbers and have not been independently audited.

DateReported figure
October 2025Deployed at 50 semiconductor companies; 50x year-over-year ARR growth; 6,377% increase in monthly usage in the first half of 2025[4]
February 2026140x year-over-year ARR growth; deployments at 80 semiconductor companies; several multi-year, multi-million-dollar licensing agreements[5]
May 2026More than 80 semiconductor companies and more than 12,000 active users[21]
July 20266x ARR growth in the first half of 2026; more than 120 semiconductor companies, including MediaTek and Micron[6]
October 2026More than 200 semiconductor enterprise deployments and a user community of more than 20,000 (website)[2]

Expanded article table

The company also says its platform cuts design cycles by more than 50%.[6] In February 2026 it cited figures from Tier-1 production programs: 15x faster specification reading, a 240x reduction in formal-assertion generation time, 400x faster UVM environment generation and 100% code and functional coverage through formal verification. Its June 2026 AWS announcement repeated these figures and described them as "internal benchmarks across customer engagements."[5][19]

Products and technology

Agent platform

ChipAgents' platform turns specifications and existing code into RTL, verification assets and automated root-cause analysis, according to the company.[11][19] Its Series A announcement described language-based commands for RTL code generation, testbench creation, debugging and verification.[4] Since the A1 round it has described its product as multi-agent "AI teams" that read specifications, break down objectives, implement and validate solutions, and iterate, rather than as a coding copilot.[5] Bessemer's Lance Co Ting Keh described it as bringing together "disparate EDA tools from spec ingestion to waveform analysis."[4] The company says its customers include architects, designers, verification and validation engineers and embedded-systems teams.[19]

Components the company has described publicly include:

  • ChipAgents RCA (November 2025), an autonomous root-cause analysis system that reads designs, testbenches, logs and waveform databases, traces a failure to its origin and suggests a fix. It uses a waveform-analysis engine built for agents and a multi-agent "prover-verifier" algorithm that tests hypotheses in parallel.[16] At a demonstration described by SemiWiki's Mike Gianfagna, the system found a bug planted in a PCIe Gen 3 design of about 50,000 lines of code in about 20 minutes, using about 20 agents. Engineers had estimated that the task would take dozens of hours by hand.[17]
  • Waveform Agents, which the company says use selective waveform extraction, cross-file signal tracing and autonomous regression triage to debug simulations whose trace data runs to terabytes.[28]
  • Generation of formal assertions, UVM verification environments and verification plans, the workflows behind the speed-up figures above.[5]

As of October 2026 the company describes its scope as autonomous design "across digital, analog, and 3D ICs."[1] It had scheduled a webinar for 29 October 2026 to unveil IC-star, which it calls its "newest autonomous chip neural engine" for end-to-end workflows and PPA (power, performance and area) convergence.[29]

Renoir model

On 17 June 2026 ChipAgents introduced Renoir, its first in-house model. The company says Renoir is fine-tuned from an open-weight mixture-of-experts language model on a curated mix of public semiconductor data and proprietary synthetic data from its own task-generation pipelines. The data covers RTL generation, specification understanding, debugging, test generation and tool use.[10] The posts do not name the base model. ChipAgents says Renoir can run fully on-premises in air-gapped environments, so customers do not have to send design files to third-party clouds.[10] On the company's internal chip-design benchmark suite, Renoir beat its base model and "approaches" Claude Opus 4.6 at less than half the cost; the June post called the release "an audition."[10] In July 2026 the company said Renoir approached Claude Fable 5 on the same internal suite at "as low as 22% of the cost."[9] These are self-reported results on a private benchmark.

NVIDIA relationship

On 26 July 2026 NVIDIA listed ChipAgents among the companies using its NVIDIA Agent Toolkit. NVIDIA said ChipAgents was building domain-specific agents for chip design and verification and fine-tuning Nemotron models for debug, formal verification, coverage and other workflows.[8] The same day ChipAgents said it was expanding its NVIDIA collaboration to train and optimize Renoir with NVIDIA Megatron Core, NeMo Megatron Bridge and NVIDIA Model Optimizer. The post quoted Kexun Zhang, ChipAgents' head of research, and Da Yang, NVIDIA's senior director of product for semiconductor and EDA.[9] ChipAgents lists NVIDIA as a strategic partner alongside Amazon Web Services and Fireworks AI, which it uses for model training and inference environments.[23]

Build-A-Chip-Bench

ChipAgents announced Build-A-Chip-Bench on 14 October 2026. The benchmark asks an agent to take a workload specification through architecture, RTL, verification, synthesis, place-and-route and signoff to a verified GDS layout, using open-source tools and an open-source process design kit.[11][12] NVIDIA contributed technical expertise to the benchmark suite and scoring rubric, and the research post also thanks Analog Devices and MediaTek.[11][12] The benchmark was developed by Surya Gunukula, Jerry Yicheng Wang and Kevin Zhang.[11] In the company's initial results, the best agent, built on Claude Fable 5.1, scored 16.6%.[12][13]

Partnerships and customers

ChipAgents joined the AWS Partner Network on 24 June 2026 and said it would use Amazon Bedrock model serving and the AWS Marketplace.[19] It said it had received a SOC 2 Type II attestation, audited by Insight Assurance, on 18 June 2026.[20] CB Insights named it to its 2026 AI 100 list in May 2026.[21]

The company has published case studies with named customers. All figures come from ChipAgents' own releases.

CustomerAnnouncedReported result
Whalechip (3D IC and wafer-on-wafer designs)15 July 2026Root-cause analysis per debug round cut from days to 15-60 minutes; four critical bugs found in a memory controller during a 60-hour debug effort[22][30]
Ambiq (ultra-low-power edge AI chips)22 July 2026Deployment expanded to more engineering teams after an evaluation[22]
Andes Technology (RISC-V processor IP)2 September 2026A processor customization task cut from 8.5 weeks to six weeks; bus interface unit verification cut from two to three months to about three weeks[22][26]
eMemory (NeoFuse NVM IP)21 September 2026Datasheet analysis and testbench construction cut from about four days to five hours (about 80%); 18% more verification test patterns[22]

Expanded article table

MediaTek and Micron are both investors and customers.[6]

Leadership

William Wang is founder, CEO and chairman of Alpha Design AI.[1] He holds the Duncan and Suzanne Mellichamp Chair in Artificial Intelligence and Design at UCSB and directs the UCSB NLP Group. He received his PhD from Carnegie Mellon University, and his UCSB homepage lists him as founder and CEO of ChipAgents.[14][15] He publishes academic work as William Yang Wang.[15][25] His honors include the IEEE Signal Processing Society's Pierre-Simon Laplace Award (2024), the British Computer Society's Karen Spärck Jones Award (2022), an NSF CAREER Award (2021), IEEE's "AI's 10 to Watch" (2020) and a DARPA Young Faculty Award (2018).[15] The company says he previously worked on Amazon Q at AWS Bedrock from 2022 to 2024.[1]

Selected executives, as listed by the company in October 2026 (Kexun Zhang's title is from a July 2026 company post):[1][9]

NameRoleBackground stated by the company
Solaiman RahimChief product officerFormer vice president of R&D at Synopsys (signoff, ECO and power platform); earlier at Atrenta and Synplicity
Mehir AroraHead of engineeringFirst employee; named a 2026 SEMICON West 20 Under 30 honoree[27]
Kexun ZhangHead of researchQuoted as head of research in July 2026[9]
Mayur SavlaHead of global salesFormerly at AWS and KLA
Cindy CuiVP, application engineering and global customer successMore than 20 years in semiconductors and EDA

Expanded article table

The board includes Wang, Wen Hsieh (Matter Venture Partners), Lance Co Ting Keh (Bessemer), Daisy Cai (B Capital), Ivan Bercovich (ScOp Venture Capital) and John Bowers. Advisers include Rhines, Camposano, Harding, Tsur, Bharathi and Luc Seraphin, president and CEO of Rambus.[1]

See also

References

  1. ^1 ^2 ^3 ^4 ^5 ^6 ^7 ^8 ^9 ^10 ^11 ^12ChipAgents, "About ChipAgents." Accessed 27 October 2026. chipagents.ai/about
  2. ^1 ^2 ^3 ^4ChipAgents, homepage ("Why ChipAgents" figures, office addresses, "© Alpha Design AI 2026"). Accessed 27 October 2026. chipagents.ai
  3. ^1 ^2 ^3 ^4 ^5Alpha Design AI (PR Newswire), "Introducing ChipAgents: the World's First AI Agent for Chip Design and Verification," via CIO Influence, 8 October 2024. cioinfluence.com/...r-chip-design-and-verification
  4. ^1 ^2 ^3 ^4 ^5 ^6 ^7 ^8 ^9ChipAgents (Business Wire), "ChipAgents Raises Oversubscribed $21M Series A to Redefine AI for Chip Design," 21 October 2025. chipagents.ai/...chipagents-series-a-21m
  5. ^1 ^2 ^3 ^4 ^5 ^6 ^7 ^8 ^9ChipAgents (Business Wire), "ChipAgents Raises $74M to Scale an Agentic AI Platform to Accelerate Chip Design," 17 February 2026. chipagents.ai/...chipagents-series-a1-74m
  6. ^1 ^2 ^3 ^4 ^5 ^6 ^7 ^8 ^9 ^10 ^11ChipAgents (Business Wire), "ChipAgents Expands Series A Funding to $134 Million as Demand Grows for Agentic AI in Semiconductor Design," 29 July 2026. chipagents.ai/...chipagents-series-a2-134m
  7. ^1 ^2 ^3 ^4Cris Tolomia, "ChipAgents was raising $60 million as it expanded its partnership with Nvidia," Quartz, 29 July 2026. qz.com/...ies-a2-funding-nvidia-partnership-072926
  8. ^NVIDIA, "NVIDIA Expands NVIDIA Agent Toolkit With NVIDIA PhysicsNeMo and CUDA-X Libraries to Transform How the World Engineers, Designs and Builds," NVIDIA Newsroom, 26 July 2026. nvidianews.nvidia.com/...ineers-designs-and-builds
  9. ^1 ^2 ^3 ^4 ^5ChipAgents, "ChipAgents Expands Collaboration with NVIDIA to Advance Agentic AI for Chip Design," ChipAgents Blog, 26 July 2026. chipagents.ai/...ia-collaboration-scale-agentic-AI
  10. ^1 ^2 ^3 ^4Tanay Biradar, Surya Gunukula, Tengxiao Liu and Kexun Zhang, "Introducing Renoir, ChipAgents' First Model," ChipAgents Blog, 17 June 2026. chipagents.ai/...introducing-renoir
  11. ^1 ^2 ^3 ^4 ^5 ^6ChipAgents, "ChipAgents Collaborates With NVIDIA to Launch Build-A-Chip-Bench for Autonomous Chip Design and Implementation," 14 October 2026. chipagents.ai/...pagents-nvidia-build-a-chip-bench
  12. ^1 ^2 ^3Surya Gunukula, Jerry Yicheng Wang and Kevin Zhang, "Introducing Build-A-Chip-Bench: from workload specification to GDS," ChipAgents Research, October 2026. chipagents.ai/...build-a-chip-bench
  13. ^Phoebe Liu, "Exclusive: Nvidia and ChipAgents Launch Evaluation For AI-Powered Chip Design," The Information, 14 October 2026. theinformation.com/...ation-ai-powered-chip-design
  14. ^1 ^2 ^3Bessemer Venture Partners, "Meet the founder and CEO of ChipAgents: William Wang," 21 October 2025. bvp.com/...nder-and-ceo-of-chipagents-william-wang
  15. ^1 ^2 ^3 ^4William Wang, faculty homepage, UC Santa Barbara Computer Science. Accessed 27 October 2026. sites.cs.ucsb.edu/~william
  16. ^1 ^2Mehir Arora and Zackary Glazewski, "ChipAgents RCA: Autonomous ASIC Root Cause Analysis," ChipAgents Blog, 5 November 2025. chipagents.ai/...chipagents-rca
  17. ^Mike Gianfagna, "I Have Seen the Future with ChipAgents Autonomous Root Cause Analysis," SemiWiki, 18 November 2025. semiwiki.com/...363689-i-have-seen-the-future
  18. ^"ChipAgents Achieves State-of-the-Art Results on NVIDIA's VerilogEval Benchmark," I-Connect007, 8 January 2025. iconnect007.com/...aep
  19. ^1 ^2 ^3 ^4ChipAgents (Business Wire), "ChipAgents and AWS Partner to Advance AI-Powered Semiconductor Engineering," 24 June 2026. chipagents.ai/...chipagents-aws-partner-network
  20. ^ChipAgents, "Trusted Agentic AI for Semiconductor Engineering: ChipAgents Achieves SOC 2 Type II Attestation," 18 June 2026. chipagents.ai/...chipagents-soc2-type-ii
  21. ^1 ^2ChipAgents, "ChipAgents Named to CB Insights' List of the 100 Most Innovative Artificial Intelligence Startups," 14 May 2026. chipagents.ai/...chipagents-cb-insights-ai-100
  22. ^1 ^2 ^3 ^4ChipAgents, "Customer Success" (Whalechip, eMemory, Andes Technology and Ambiq case studies). Accessed 27 October 2026. chipagents.ai/customers
  23. ^ChipAgents, "Partners." Accessed 27 October 2026. chipagents.ai/partners
  24. ^1 ^2ChipAgents, "Building the Future of AI-Driven Semiconductor Design: Celebrating 2 Years of ChipAgents," ChipAgents Blog, 11 June 2026. chipagents.ai/...celebrating-2-years-of-chipagents
  25. ^1 ^2Wenhan Xiong, Thien Hoang and William Yang Wang, "DeepPath: A Reinforcement Learning Method for Knowledge Graph Reasoning," arXiv:1707.06690, 2017. arxiv.org/...1707.06690
  26. ^ChipAgents (Business Wire), "ChipAgents Helps Andes Technology Dramatically Accelerate Custom Processor Design and Verification," 2 September 2026. chipagents.ai/...andes-technology-custom-processor
  27. ^1 ^2ChipAgents, "ChipAgents' Mehir Arora named a 2026 SEMICON West 20 Under 30 honoree," 12 October 2026. chipagents.ai/...ir-arora-semicon-west-20-under-30
  28. ^ChipAgents, "One Year In: Taming the Waveform Tsunami with AI Agents," ChipAgents Blog, 28 May 2026. chipagents.ai/...taming-waveform-tsunami
  29. ^ChipAgents, "Unveiling IC-star" (webinar page). Accessed 27 October 2026. chipagents.ai/...ic-star-webinar-2026
  30. ^ChipAgents (Business Wire), "ChipAgents Helps Whalechip Cut Semiconductor Root Cause Analysis from Days to Minutes," 15 July 2026. chipagents.ai/...chipagents-whalechip-rca

Improve this article

Add missing citations, update stale details, or suggest a clearer explanation. Every suggestion is reviewed for sourcing before it goes live.

1 revision · v1 · 2,770 words · full history

Fact-checks are independent of edits: a reviewer re-verifies the article against its sources and stamps the date. How we verify

Research and drafting on this wiki are AI-assisted, under named human editorial standards. How AI is used here

Reviewer note: Independent adversarial verification 2026-10-27 (xg23): claims checked against primary sources; verifier fixes applied

Cite this page: AI Wiki. "ChipAgents." aiwiki.ai, updated 27 Oct 2026, fact-checked 27 Oct 2026. CC BY 4.0. https://aiwiki.ai/wiki/chipagents

Suggest edit

Where to go next

Topics:AI AgentsAI CompaniesAI Hardware