Ondine Capital Fund III Reaches First Close for AI Bets

The semiconductor companies backing Ondine Capital Fund III occupy a useful place in the AI economy: they help supply the machinery other startups need to make their promises work. Bringing those businesses into a venture fund creates a relationship worth examining as Ondine invests across models, chips, infrastructure and applications.

The firm announced the first close of its third USD-denominated venture fund on November 3, 2026, with institutional and strategic limited partners, including unnamed Taiwanese semiconductor enterprises. It also appointed Brian Yen as partner, focused on early-stage intelligent software and physical AI. Neither the first-close amount nor a target size was disclosed.

Ondine's early Fund III portfolio includes Recursive Superintelligence, Salience Labs and Bot Auto. Those investments make the mandate tangible: software that helps conduct AI research, hardware that moves data through computing systems, and intelligence deployed in freight transportation. The commercial work changes substantially between those settings, giving the firm's industrial relationships a different job in each one.

A semiconductor network meets a broader AI mandate

Ondine says Fund III will invest across foundation models, next-generation semiconductors, AI infrastructure, AI-native software and physical AI. That is a wide remit for any investment team. The useful question for a founder is which parts of the network can help with the particular technical or commercial decision in front of the business.

A semiconductor company can bring knowledge of production and system requirements to an investment conversation. Whether that knowledge becomes useful depends on people making time, understanding the startup's problem and finding a practical reason to collaborate. An LP commitment establishes a financial relationship; the announcement provides no evidence of promised purchase contracts or privileged manufacturing access.

The distinction matters because industrial participation is easy to oversell. A founder trying to qualify a component, recruit a specialist or understand a customer's roadmap needs a specific answer from somebody responsible for that work. An investor's general familiarity with the sector may help start the conversation, while the company still has to earn the commercial relationship.

Ondine's official firm profile describes early-to-growth investing across East Asia and the US, with a presence in Taipei, Singapore, Tokyo and Silicon Valley. That geography fits a strategy reaching between engineering teams and markets. Its usefulness to a particular portfolio company will depend on the people Ondine can put into the same discussion.

Optical switching gives the strategy a concrete example

Salience Labs offers a clearer view of the work behind the infrastructure label. The Oxford company announced a $125M Series B on October 14, led by Bullhound Capital, with AMD Ventures and Intel Capital among the participants. Salience said the financing would support optical circuit switching deployment, manufacturing scale and expansion of its US presence.

Ondine separately confirmed its participation in that round. Salience's funding belongs to the portfolio company and cannot be used as an estimate of Fund III's size. The investment is useful here because it connects Ondine's stated mandate to a product moving toward larger-scale production.

Salience develops silicon-photonics switching technology for AI data centers. As the company describes its architecture, optical switches connect GPUs across racks, making communication between computing resources part of the performance problem. Financing that business brings an investor close to manufacturing readiness and infrastructure design, alongside the familiar questions about growth and customer demand.

The portfolio also reaches into a different part of computing. Ondine's October update on Recursive Superintelligence describes Flint, a local coding model developed using the company's Eureka Machine automated research system. Those are company-reported developments, and no benchmark superiority is assumed here. Alongside Bot Auto's autonomous-freight focus, they illustrate how many different forms of technical judgment sit inside an apparently simple AI investment mandate.

Brian Yen adds a founder-facing role to the fund

Brian Yen's appointment gives the first-close announcement a named operating change. Ondine says he previously served as principal and head of the Taipei office at Headline Asia, developing relationships across Japan, Taiwan and Southeast Asia. At Ondine, his focus will be early-stage founders building intelligent software and physical AI platforms.

The firm dates its founding to 2019. Co-founder Randolph Hsu and founding and managing partner Yen An Cho remain part of its leadership identity, while Brian Yen's arrival adds another investor responsible for finding and supporting founders. The expanded investment scope asks that leadership group to stay useful across very different development cycles.

For a software founder, an infrastructure specialist and a transportation team, support will mean different things. An introduction has value when the person taking the meeting understands the decision the startup needs to make. Fund III's combination of industrial backers, a broad AI portfolio and Brian Yen's early-stage remit gives Ondine several places to begin that work.

The first close leaves the vehicle's eventual scale open. In the meantime, Ondine is already presenting companies whose progress runs through research, production and deployment. Brian Yen joins a team with those relationships in motion, and founders who will need an answer more specific than an investor's enthusiasm for AI.