A San Francisco startup that promises to design custom computer chips from simple text prompts has secured $5.2 million in seed funding, led by Uncork Capital with backing from Moxxie Ventures and angel investor Jeff Dean. Phinity Labs, which emerged from a pre-seed round led by Pear's PearX program, claims it can already count some of the largest AI research labs among its early customers.
The company says it has reached an eight-figure annualized run rate. Amy Saper, a general partner at Uncork who joined Phinity's board, posted on LinkedIn that the startup hit that milestone in under three months. These figures are company-reported and have not been independently verified, and Phinity has not publicly disclosed its customers or valuation.
What Phinity is building represents an ambitious bet on automation in one of tech's most labor-intensive fields. The startup's software agents handle the entire chip design process: architectural planning, verification testing, and final tapeout preparation. According to the company, its goal is to deliver production-ready GDS files in weeks rather than the months or years typical today, targeting that timeline by 2028.
Phinity frames the technology as a closed-loop system where AI agents design a chip, evaluate its performance against specifications, then iterate based on physical feedback. The company claims its system produces verified designs significantly faster than experienced hardware engineers using current-generation coding tools, while achieving better metrics for power consumption, performance, and die area. Those benchmarks, too, remain unverified by outside parties.
A team drawn from chip giants
Cofounder Sonya Jin previously developed synthetic data pipelines for NVIDIA's chip design tools and worked on code-reasoning language models, in addition to conducting graph machine learning research at Amazon's AI Lab. Her cofounder, Aadi Nashikkar, is a repeat founder with experience in synthetic data generation and model post-training. LinkedIn profiles suggest the company employs roughly eight people, including engineers Albert Chun and Kunwoo Min.
Phinity says its roster includes a former leader from Intel's custom silicon division, an ASIC lead from Kepler Computing, and NVIDIA architects who managed multiple chip tapeouts on cutting-edge manufacturing nodes. The company also highlights AI researchers who have trained language models specifically for chip design and formal verification tasks, though it did not provide names for these hires.
Racing against established players
Synopsys and Cadence, the two companies that dominate electronic design automation software, have spent years integrating AI into their toolchains. Synopsys announced its first 100 commercial tapeouts using DSO.ai in early 2023, roughly four years after unveiling the technology. Cadence introduced Cerebrus Intelligent Chip Explorer, a reinforcement-learning-based optimization tool, in mid-2021.
Academic researchers have accelerated work on agentic chip design in recent months. Preprints exploring the concept include "AiEDA: Agentic AI Design Framework for Digital ASIC System Design" and "ASIC-Agent: An Autonomous Multi-Agent System for ASIC Design with Benchmark Evaluation." Other research has examined natural-language interfaces for open-source chip design workflows, though the practical applicability of these approaches at commercial scale remains an open question.
Phinity plans to use the seed capital to grow its engineering team and scale the infrastructure behind its agent systems. The company is actively recruiting for roles in chip design and AI systems engineering, according to a careers page it hosts on Ashby.
The funding arrives as AI companies face mounting pressure to secure dedicated computing resources. Custom chips tailored for specific workloads offer potential advantages over general-purpose processors, but the design process has historically required deep expertise and lengthy iteration cycles. Whether software agents can compress that timeline without sacrificing reliability will determine if Phinity's approach gains traction beyond early adopters.
