Ricursive

Ricursive Intelligence is an American artificial intelligence research lab founded in 2025 in Palo Alto, California, by Anna Goldie and Azalia Mirhoseini, the researchers who co-created AlphaChip at Google DeepMind. The company develops frontier AI methods for semiconductor design and is building toward a closed loop in which AI models design the next generation of chips and those chips, in turn, train more advanced models. As of May 2026, Ricursive has raised approximately $335 million across seed and Series A rounds at a $4 billion post-money valuation, four months after its public launch.

At a glance

• Founded: 2025 in Palo Alto, California, by Anna Goldie and Azalia Mirhoseini.
• Status: Private. Series A closed February 2026.
• Funding: Approximately $335 million cumulative. $35 million seed in December 2025 led by Sequoia Capital at a $750 million post-money valuation. $300 million Series A in February 2026 led by Lightspeed at a $4 billion post-money valuation, with DST Global and NVIDIA's NVentures participating.
• CEO: Anna Goldie, co-founder. Former Senior Staff Research Scientist at Google DeepMind and an early researcher at Anthropic, where she contributed to Gemini and Claude.
• Other notable leadership: Azalia Mirhoseini, co-founder. Assistant Professor at Stanford University, where she runs the Scaling Intelligence Lab. Previously a Senior Staff Scientist at Google DeepMind and Anthropic.
• Open weights: None disclosed. Ricursive's commercial output is enterprise design tooling rather than model releases.
• Flagship outputs: AlphaChip (Nature 2021, predecessor research at Google), RL-CCD (DAC Best Paper 2023), Insta (DAC Best Paper 2025), and C3PO (ASP-DAC Best Paper 2026). Public productization roadmap is undisclosed.

Origins

Goldie and Mirhoseini have been collaborators for nine years, dating to a single day at Google Brain in 2016 when both joined the same team. Their AlphaChip project, published in Nature in 2021, applied deep reinforcement learning to the floorplanning problem in chip design, generating layouts that outperformed human-designed baselines. Google deployed AlphaChip in four generations of TPU and across additional internal silicon, and external chipmakers subsequently adopted derivative methods.

Goldie completed her PhD in the Stanford NLP Group and holds undergraduate and master's degrees from MIT. Mirhoseini received her PhD from Rice University, where her dissertation won a best-thesis award. Both researchers spent time at Google DeepMind and at Anthropic, where they contributed to flagship language model lines, before regrouping around the chip-design thesis that had defined their joint work at Google Brain.

The company emerged from stealth on December 2, 2025, with the announcement of a $35 million seed round led by Sequoia at a $750 million valuation, an unusually rich valuation for a pre-product seed and an indicator of the perceived strategic importance of AI-for-chip-design as a category. Two months later, in February 2026, Lightspeed led a $300 million Series A at a $4 billion post-money valuation, with DST Global and NVIDIA's NVentures participating. The four-month interval from launch to a $4 billion valuation is among the fastest such trajectories in recent AI venture history.

The company's stated team draws senior researchers from Google DeepMind, Anthropic, NVIDIA, Cadence, Apple, xAI, Stanford, MIT, and Harvard. Mirhoseini continues her Stanford professorship in parallel with her co-founder role.

Mission and strategy

Ricursive's stated mission is "recursive self-improvement via AI for chip design and chip design for AI." The framing reflects a specific strategic premise: that AI-driven semiconductor design is a leverage point for accelerating broader AI capability progress, because better chips train better models and better models can design better chips. The company positions itself as a frontier AI lab whose unit of progress is the design loop rather than the model release.

The strategy combines three threads. The first is foundation-model research applied to the unique structure of semiconductor design, including floorplanning, placement and routing, datapath optimization, and verification. The second is a productization track for enterprise chipmakers and EDA customers, where the commercial product is design tooling that compresses physical-design cycles. The third is internal use of the same tooling on Ricursive's own AI infrastructure, creating the recursive feedback loop that gives the company its name.

The competitive premise is that AI for chip design is structurally different from horizontal AI applications. The dataset, the constraints, and the evaluation criteria are specialized; general-purpose foundation models can contribute, but a focused lab with deep semiconductor expertise can produce designs that outperform general-purpose approaches. The premise also assumes that incumbent EDA vendors (Synopsys, Cadence, Siemens) will face structural disadvantages adapting their existing toolchains to AI-native design methodologies.

Models and products

• Foundation-model research for chip design. Ricursive's primary research output is methodology rather than model checkpoints. The company has continued the AlphaChip research line with subsequent papers including RL-CCD (DAC Best Paper 2023), Insta (DAC Best Paper 2025), and C3PO (ASP-DAC Best Paper 2026).
• Enterprise design tooling. Productization details are largely undisclosed as of May 2026. Industry coverage has indicated that the commercial product is design tooling for chipmakers, with revenue and pilot-customer details not publicly broken out.
• Internal-loop deployment. The "recursive" framing implies internal use of Ricursive's tooling on the company's own compute fleet, though the operational specifics are also undisclosed.

Distribution is positioned for direct enterprise sales to chipmakers and EDA customers, with no consumer-facing product. The company has not announced API access or open-weights releases.

Benchmarks and standing

Ricursive's evaluation framework is design-specific rather than horizontal. The relevant benchmarks are physical-design metrics (placement quality, wirelength, timing closure, area utilization) and the cycle-time compression realized by AI-assisted versus human-assisted design flows. The company has not published comparative leaderboards against incumbent EDA tools, so external benchmarking depends on academic papers from the founders' research lineage and on customer-reported case studies once those become available.

Industry coverage has frequently characterized Ricursive as the leading AI-native chip-design lab globally based on the founder team's track record on AlphaChip and subsequent best-paper outputs at DAC and ASP-DAC, the principal academic venues for electronic design automation. Whether the productization can convert that research lead into commercial advantage against the incumbent EDA stack is the central open question.

Leadership

As of May 2026, Ricursive's senior leadership includes:

• Anna Goldie, co-founder and Chief Executive Officer. Co-led the AlphaChip project at Google Brain. Senior research scientist at Google DeepMind and an early researcher at Anthropic, with contributions to Gemini and Claude.
• Azalia Mirhoseini, co-founder. Assistant Professor at Stanford University running the Scaling Intelligence Lab. Co-led AlphaChip and subsequent AI-for-chip-design research. Senior Staff Scientist roles at Google DeepMind and Anthropic.

The senior research and engineering bench is reported to include hires from Google DeepMind, Anthropic, NVIDIA, Cadence, Apple, xAI, and several major research universities. Specific senior-leadership titles beyond the co-founders have not been publicly disclosed.

Funding and backers

Ricursive's funding history through May 2026:

• Series A (February 2026): $300 million at $4 billion post-money valuation, led by Lightspeed with DST Global and NVIDIA's NVentures participating.
• Seed (December 2025): $35 million at $750 million post-money valuation, led by Sequoia Capital.

Cumulative reported equity is approximately $335 million. NVIDIA's strategic-investor participation through NVentures is notable given NVIDIA's position as both a customer and a competitor in AI-for-chip-design tooling. Sequoia, Lightspeed, and DST Global are generalist growth investors with track records in foundational AI infrastructure.

The Series A close at a $4 billion post-money less than two months after public launch reflects investor demand for AI-for-chip-design exposure and the founder team's perceived scarcity premium. Whether the valuation is supported by realized commercial metrics is among the open questions for 2026 and 2027.

Industry position

Ricursive occupies a distinctive position as an AI-native chip-design lab founded by the principal researchers behind AlphaChip. The combination of the founder team's research lineage, the strategic-investor support including NVIDIA's NVentures, and the recursive-design framing produces a profile that the incumbent EDA vendors and the foundation-model labs do not directly match.

Industry coverage has characterized the company as a leverage point for AI-driven semiconductor design, with the founder team identified as the principal architects of the AlphaChip research line. Analysts have also flagged execution risk: a $4 billion valuation implies aggressive expectations for commercial conversion, and the EDA market has structural characteristics (long sales cycles, deep customer integration, certification requirements) that historically reward incumbents.

Competitive landscape

Ricursive's competitive landscape spans multiple categories:

• Incumbent EDA vendors. Synopsys, Cadence, and Siemens EDA are the three principal incumbents. Each has its own AI-augmented design products, though the underlying toolchains were built before the foundation-model era. Ricursive's premise is that an AI-native architecture compresses cycles in ways the incumbents will struggle to match without rebuilding their stacks.
• Foundation-model labs with chip-design components. Google DeepMind continues AlphaChip-line research internally. NVIDIA's research arm and other major labs (Microsoft AI, Amazon AGI, Apple) have internal AI-for-design programs.
• AI hardware startups. Groq, Cerebras, SambaNova, and others compete for AI-infrastructure share but do not directly compete in chip-design tooling.
• Specialized AI-for-EDA startups. A small but growing set of startups including ChipFlow, Astrus, and ChipAgents target subsets of the chip-design workflow with AI methods. Ricursive's positioning is broader.

Outlook

Several open questions affect Ricursive's trajectory in 2026 and 2027:

• Productization timeline and the first publicly announced enterprise customers.
• Realized revenue against the $4 billion valuation, with Series B financing as the next pricing event.
• The scope of the recursive-design loop in practice, including whether internal use produces measurable capability gains over off-the-shelf approaches.
• The competitive response from Synopsys and Cadence, including any acquisitions of AI-for-EDA startups or partnerships that could narrow Ricursive's positioning.
• Senior research and engineering hiring, given the demand for AI-and-EDA crossover talent.
• Continued academic-paper output and the conversion of those research milestones into commercial product features.

Sources

• Sequoia Capital: Partnering with Ricursive Intelligence. Lead-investor announcement at seed.
• TechCrunch: How Ricursive Intelligence raised $335M at a $4B valuation in 4 months. Series A reporting.
• Crunchbase News: Ricursive Series A at $4B. Series A coverage.
• Data Center Dynamics: Ricursive Intelligence raises $300m against $4bn valuation. Funding context.
• Yahoo Finance: Ricursive Intelligence Launches Frontier AI Lab. Launch announcement.
• Google DeepMind: How AlphaChip transformed computer chip design. Predecessor research lineage.
• Ricursive official site. Company reference.