Participated · Series A · Aug 2026

Vals AI builds an AI evaluation platform that works with domain experts to convert real-world workflows into rigorous benchmarks and then creates automated grading systems to assess final work products to expert standards. The company positions its product around expert-designed evaluation and benchmarking rather than model training. As of the reported financing, Vals AI raised $40 million in a Series A at a $400 million valuation. Existing investors 8VC, Pear VC, and Bloomberg participated in the round alongside new backers Hudson River Trading and NextLadder Ventures, with Andreessen Horowitz leading. The articles do not report revenue, user counts, or specific go-to-market metrics, nor do they disclose explicit use-of-proceeds details.

Participated · Series B · Aug 2026

Founded in London by James Dacombe in 2024, Olix develops the X-1 platform that distributes AI model workloads across specialised chips connected by an optical interconnect. Its first chip, the DX-1, is a decode accelerator that the company claims can deliver over 10,000 tokens per second per user on 100-billion-parameter models while using less power for large batches, and is designed to scale to models with trillions of parameters. Olix has over 140 employees across London, Bristol, Toronto, Austin and San Francisco. The company plans to launch DX-1 to initial customers in the second half of 2027 and to ramp platform development and chip manufacturing with the new funding. Olix has rapidly increased its valuation through multiple funding rounds, most recently reaching $3.3 billion after the $312 million raise, up from just over $1 billion in February. The AI chip market in which Olix competes was valued at about $84 billion in 2026 and is expected to grow rapidly through 2030.

Participated · Series A · Jun 2026

Gray Swan builds an AI security platform that combines real-time protection (Cygnal), automated adversarial testing/red-teaming (Shade), and a global red-teaming community (Arena) to detect and prevent dangerous model and agent behavior. Cygnal enforces custom enterprise policies in real time, while Shade runs automated pre-deployment adversarial tests as part of CI/CD, and Arena crowdsources attacks from more than 15,000 researchers to generate over one million attack trajectories. The company says its benchmarks are embedded in the safety evaluation processes of leading frontier labs and that it is cited in 11 recent frontier model system cards, including those from Anthropic, OpenAI, and Meta. Gray Swan works with over 20 customers and has partnerships with platforms such as Snowflake, including a native integration for runtime AI protection. The recent $40M Series A will fund go-to-market scaling, deeper lab and platform partnerships, and team growth to support enterprise deployments.

Participated · Series A · Mar 2026

Founded in 2024 by serial networking entrepreneur Drew Perkins and chip designer Omar Hassen, Eridu is redesigning data-center networking from the silicon up to meet the bandwidth demands of modern AI workloads. The company is building a switch that integrates far more networking functionality directly onto the chip, reducing optical links, lowering latency, and cutting power and cost compared with traditional multi-tiered architectures. Its eventual product line will replace conventional gear from vendors like Arista Networks in AI data centers, positioning Eridu at the heart of the ongoing build-out of GPU clusters that now span millions of processors. Perkins argues that while GPU compute and memory bandwidth are improving roughly 10× per year, incumbent switch technology advances only 2–3× every few years, creating a performance gap Eridu aims to close. The startup has assembled roughly 100 employees and leverages the founders’ deep experience, including Perkins’ past exits such as Lightera, Infinera, and Gainspeed. Although the company has not disclosed revenue metrics or valuation, it has raised substantial capital to fund product development and scale.

Participated · Series B · Dec 2025

Etched.ai develops rack-scale inference systems and broader frontier inference infrastructure. Since founding, the company has grown to more than 400 employees, hiring over 200 people after its Series B late last year. It has secured more than $1 billion in customer orders and established deep manufacturing and supply-chain partnerships, including opening a factory in Taiwan. Etched is currently validating its first rack-scale inference product with customers. The company’s near-term plans emphasize production, manufacturing scale-up, and deployment of its inference systems, supported by cumulative funding of $800 million.