A Firm That Keeps Raising More Money — and Keeps Earning It
Most VC firms launched in 2022 are having a rough few years. Capital is scarce, LP patience is thin, and a lot of emerging managers who rode the ZARA era into existence are now quietly shutting down or limping toward a final close. Dimension Capital is not one of those firms.
Founded in late 2022 by Zavain Dar, Adam Goulburn, and Nan Li, the New York firm just closed an $800 million third fund — announced Tuesday — representing a 60% step-up from the $500 million second vehicle it raised just 18 months earlier. That's not a fundraising arc you see often, and it says something real about how limited partners are reading the deep-tech moment right now.
Dar and Goulburn came to Dimension from Lux Capital, one of the more respected science-focused shops in the industry. Li joined from Obvious Ventures. When they launched the firm, their core bet was blunt: founders would increasingly want to build companies that dissolved the old wall between biotech and software. Compute would eat biology the same way it had eaten everything else. Turns out they were right — and they got there early enough to matter.
The $800M Third Fund: What It Says About the Market
The headline number is $800 million. That's big — but the more interesting number is the velocity. Dimension went from fund two to fund three in 18 months, which is fast by any standard. You don't do that unless your portfolio is performing and your LPs want more exposure.
The fund's thesis stays centered on what the firm calls the intersection of science and compute. That phrase gets thrown around a lot, but Dimension means something specific: companies where the scientific breakthrough and the computational infrastructure are genuinely inseparable. Not "biotech with an ML layer." Companies where neither half could exist without the other.
From a market positioning standpoint, this fund lands at a moment when the broader AI investment landscape is stratifying fast. Some funds are piling into pure-play software and AI developer tools; others are moving toward physical infrastructure. Dimension is betting that the most durable value will be created at the seam — in companies that can only be built when you have both scientific insight and serious compute. It's a view that's gaining traction globally, including in markets like India, where AI developer tools startups and infrastructure investments are increasingly flowing to firms that can show cross-domain depth, not just software chops.
Chai Discovery: The Seed Bet That Returned the Thesis
If you want to understand how Dimension's playbook works in practice, start with Chai Discovery.
In 2024, the firm co-led a $30 million seed round in Chai Discovery, a startup building open-source AI foundation models for drug development. That's a meaningful seed check for a science-heavy company — the kind of bet that requires real conviction in both the biology and the infrastructure needed to make it work.
Last week, Chai Discovery announced it had raised $400 million at a $3.8 billion valuation. That's the kind of outcome that makes LPs take a second look at a fund's track record on paper. It also validates something specific about the approach: betting on open-source AI infrastructure for drug discovery, at the seed stage, before most of the industry was paying attention to that corner.
The open-source angle matters here. Chai isn't trying to lock up proprietary models behind a wall — it's betting that the scientific community's adoption of open tooling drives network effects that ultimately feed a commercial platform. That's a software-native strategy applied to a deeply scientific problem. Exactly the kind of company Dimension was built to back.
New Limit and the Anti-Aging Bet Paying Off
New Limit is the other marquee name in this fund cycle, and it's a slightly stranger story.
The anti-aging startup is co-founded by Brian Armstrong — yes, the CEO of Coinbase — and it's been quietly building toward a position that, until recently, only a handful of investors believed was fundable at scale. Dimension backed New Limit at its Series A in January 2025. Earlier this month, the company closed a Series C at a $3.1 billion valuation.
That's a rapid climb from Series A to a $3.1 billion Series C in roughly 18 months. It also reflects something broader happening in longevity and anti-aging science: capital is finally taking the category seriously as a place where biological insight and computational modeling are genuinely converging. New Limit is less a pharma bet and more a deep-tech platform play — exactly the kind of framing Dimension is comfortable with.
Armstrong's involvement as a co-founder brings a specific kind of credibility to the table. He's not a scientist, but he's a builder who has scaled complex technical infrastructure, and the company has attracted serious scientific co-founders alongside him. The $3.1 billion valuation suggests the market is buying the thesis.
Modal Labs, Anthropic, and the Inference Layer
Beyond the two marquee names, Dimension's portfolio tells a consistent story about where the partners think value accumulates in the science-compute stack.
Modal Labs is an inference company — infrastructure for running AI workloads efficiently at scale. It's not a biotech play. But it's a foundational piece of the compute side of Dimension's thesis: if you're going to run AI models for drug discovery or aging research, you need inference infrastructure that can handle the workload without burning through a fortune in GPU costs. Modal is a bet on that layer.
The Anthropic connection is more indirect. Dimension received shares in Anthropic after the AI giant acquired Coefficient Bio — a drug discovery platform that was in Dimension's portfolio — for a reported $400 million this spring. That's a meaningful outcome for the fund, and it's a useful illustration of how the science-compute crossover is playing out at the institutional level: the biggest AI labs are actively acquiring scientific application companies, which creates a natural exit path for deep-tech investors who got in early.
For context on the broader AI investment landscape this fund is operating in, Greylock's measured $1.5B raise offers a useful contrast — a different philosophy about fund sizing at a similar moment. And the infrastructure bets running in parallel, including the move from LLM wrappers to physical infrastructure, shape the macro environment Dimension is navigating.
Why This Matters Beyond the Numbers
The $800 million headline is worth unpacking from a few different angles.
First, it's a signal about LP appetite for science-focused funds specifically. General AI is crowded; pure biotech is struggling with its own set of headwinds. But deep-tech firms that can credibly straddle both — and show portfolio results to back it up — are finding real demand. Dimension's rapid step-up from fund two to fund three suggests LPs see this as a differentiated position, not another AI generalist.
Second, it matters for the companies getting built in this space. An $800 million fund can write meaningful checks at the seed and Series A stage while holding meaningful reserves for follow-on. That's relevant for founders building at the science-compute intersection, where the capital requirements tend to be higher and the development timelines longer than pure software. Having a well-capitalized, thesis-consistent lead investor who will follow on matters a lot when you're three years into a hard biology problem.
Third, it's a data point about the broader competitive landscape for AI developer tools startups and infrastructure bets globally — including in markets like India, where AI and compute investments are accelerating rapidly, and where the appetite for cross-domain deep-tech is growing alongside pure software plays. HCL's move into AI datacenters, for instance, reflects exactly this appetite for full-stack AI infrastructure in India. The kind of thesis Dimension is running isn't geography-bound; the science-compute crossover is a global opportunity, and the firms that built early track records in it are now raising at scale.
The Three Partners and the Bet They Made
It's worth spending a moment on the founding team, because the thesis didn't emerge from nowhere.
Zavain Dar and Adam Goulburn both came from Lux Capital, which has spent over two decades investing in hard science companies — robotics, synthetic biology, materials science. That background wires you to think about long development cycles, proprietary scientific advantage, and the specific kinds of risk that biotech companies carry. It also means you're comfortable with companies that can't ship an MVP in six weeks.
Nan Li's background at Obvious Ventures brings a different flavor — a fund that focused on what it called "world positive" investing, often at the intersection of technology and sustainability. Between the three, you get a founding team that's comfortable with complexity, long timelines, and category-creation risk.
When they launched in late 2022, the environment was not particularly favorable for that combination. The market correction was in full swing, deep-tech was being questioned as an investable category, and plenty of LPs were pulling back from anything that wasn't pure software. The fact that they've now raised three funds in roughly four years — each significantly larger than the last — is a meaningful vindication of the decision to launch when they did.
The partners themselves admit to being surprised by how quickly their thesis has played out. That's not false modesty — the pace at which companies like Chai Discovery and New Limit have reached multi-billion dollar valuations is genuinely faster than most deep-tech predictions would have suggested even three years ago. Compute is moving faster than the science, and that's pulling the whole category forward.
What Comes Next
An $800 million fund buys Dimension time and firepower to extend its portfolio into the next generation of science-compute founders. The firm will likely continue co-leading seed and Series A rounds in companies that look like early Chai or early New Limit — technically ambitious, open to the compute-native approach to scientific problems, and building in categories that most generalist VCs still don't fully understand.
The Anthropic acquisition of Coefficient Bio is also a template worth noting. As the largest AI labs get serious about scientific applications — drug discovery, materials science, climate modeling — they're going to keep acquiring companies in these spaces. That creates a clear exit path for early investors in the category. Dimension, having already seen that outcome once, will almost certainly structure future investments with that dynamic in mind.
For founders building at the science-compute seam, and for the LPs trying to decide who earns their next check, Dimension's trajectory over the last four years is a clear statement of position. The intersection of science and compute isn't a niche. It might be where the next generation of category-defining companies gets built.
Source
- Dimension Capital, founded in late 2022, announced an $800 million third fund on July 21, 2026
- The fund is 60% larger than its $500 million second vehicle announced just 18 months prior
- Founders: Zavain Dar and Adam Goulburn (formerly partners at Lux Capital), and Nan Li (Obvious Ventures alum)
- Thesis: founders would increasingly want to build deep-tech companies crossing boundaries of biotech and software
- In 2024, Dimension co-led a $30 million seed investment in Chai Discovery
- Chai Discovery builds open-source AI foundation models for drug development; raised $400 million at a $3.8 billion valuation
- Dimension backed anti-aging startup New Limit at its Series A in January 2025
- New Limit, co-founded by Coinbase CEO Brian Armstrong, completed a Series C at a $3.1 billion valuation in July 2026
- Other notable investments include inference company Modal Labs and Anthropic
- Dimension received shares in Anthropic after Anthropic acquired portfolio company Coefficient Bio for a reported $400 million