Gradient Ventures Closes $220M Fifth Fund to Seed the Next Wave of AI Founders
TL;DR
Gradient Ventures has closed its fifth fund at $220 million, marking the firm's first raise with outside institutional limited partners since spinning out of Google's Alphabet last year. Founded in 2017 as Google's in-house AI seed vehicle, the San Francisco-based firm now manages nearly $1.2 billion across five funds and has backed more than 500 AI companies — with exits to NVIDIA, Snowflake, and Axon already on the books. This raise is the clearest signal yet that Gradient has successfully made the transition from corporate venture arm to independent fund.
Key Takeaways
Independence from Google is now real, not just nominal. For the first four funds, Google was the sole LP. Fund V changes that structural reality: Alea Capital led the LP syndicate, joined by Reflex Capital, Smartlink, and J&T Ventures. Google remains in, but as one voice among several. That shift in governance and incentive alignment matters more than the headline number.
The deal flow flywheel is Gradient's real moat. When ChatGPT launched, Gradient's inbound company pipeline exploded from roughly 100 submissions per year to somewhere between 1,500 and 2,000. Seeing that volume of early AI companies — and having seven years of pattern recognition on what actually works — creates a sourcing advantage that most funds cannot replicate.
Gradient is disciplined about what it won't fund. The firm explicitly avoids foundational model companies and has expressed skepticism toward mega-seed rounds above $100 million. At a time when every VC seems to be writing $50M+ seed checks, that restraint is either principled or a competitive blind spot. The exit track record suggests it's principled.
The exit record is genuinely impressive for a seed-stage firm. CentML was acquired by NVIDIA at a reported $400M+, Streamlit went to Snowflake for $700-800M, and Prepared was acquired by Axon. For a fund that started when AI was still considered a niche, those exits validate the thesis long before the current AI investment mania arrived.
Why This Fund Matters
Gradient was purpose-built for AI investing before AI investing existed as a category. Google launched the firm in 2017 — one month after the release of the "Attention Is All You Need" paper that would eventually underpin the transformer architecture behind every major language model today. At the time, that timing looked early to the point of strange. In hindsight, it was one of the better-timed institutional bets in venture history.
What makes the Fund V close more significant than the dollar figure is the structural transformation it represents. Gradient is no longer a corporate venture arm operating under Alphabet's balance sheet. Management company ownership has transferred to the founding partners, Darian Shirazi and Zach Bratun-Glennon, and the LP base now includes independent institutions who will hold the firm accountable to market-rate returns rather than Google's strategic priorities. That transition from CVC to independent GP is notoriously hard to execute — most attempts fail either at fundraising or at retaining talent. Gradient appears to have navigated both.
The fund's sector stance is worth understanding carefully. Gradient has no interest in funding foundational model labs, which is a deliberate departure from the trajectory most high-profile AI funds are pursuing. The firm's thesis is that the application and agentic layer — companies building on top of models rather than building models themselves — is where durable value will accrete. That thesis was controversial two years ago. The performance of portfolio companies like Writer, Lambda, and Krea suggests it is aging well.
At $220 million, this fund size is notable for what it isn't: it is not another $1 billion mega-seed vehicle. Gradient is scaling its fund size modestly relative to AUM, keeping check sizes calibrated to genuine seed rounds rather than ballooning into a growth fund in seed clothing. In a market where "seed" now routinely means $20-40 million rounds with institutional term sheets, that restraint keeps Gradient in the market segment where it has actually demonstrated alpha.
The Team
Darian Shirazi and Zach Bratun-Glennon are both engineers-turned-investors, which shapes how they engage with portfolio companies. Shirazi joined Google to help incubate what became Gradient and has spent nearly a decade building the firm's thesis around AI companies that generate real-world utility, not benchmark performance. Bratun-Glennon has been vocal about his conviction that we are in "the largest platform shift in history" and "the biggest value creation event in technology ever" — a view that reads as promotional in many mouths but is somewhat more credible coming from someone who has been making seed bets on AI companies since 2017.
Early Portfolio
Gradient has backed more than 500 companies across its five funds. The current active portfolio includes Writer (enterprise generative AI), Lambda (AI cloud infrastructure), Krea (AI-powered creative tools), Oura (wearable health intelligence), Sona (AI-native workforce management), and Airspace Intelligence (AI for aviation logistics). Notable exits include CentML (acquired by NVIDIA for a reported $400M+), Streamlit (acquired by Snowflake for $700-800M), and Prepared (acquired by Axon). The depth and range of that portfolio — covering infrastructure, applications, health, and enterprise — reflects a genuinely thesis-driven approach rather than sector spray.
What This Means for Founders
For founders building at the AI application or agentic layer, Gradient is now a more interesting conversation than it was a year ago. With independent institutional backing, the firm has less reason to steer portfolio companies toward Google's strategic interests and more reason to optimize for exits and returns. If your company is building on top of models — not competing with them — and you are at pre-seed or seed stage, Gradient's combination of deal flow volume, AI-specific operator network, and genuine exit track record makes it one of the better-positioned seed funds in the market.
What Gradient is not is a fit for foundational model companies, defense-tech pure plays, or founders seeking a mega-seed round north of $50 million. The firm is disciplined about its lane. Founders who have raised $100M pre-launch should look elsewhere. Founders who want a partner with seven years of AI pattern recognition and a portfolio of companies who have successfully navigated the application layer should pay close attention to Gradient Fund V's deployment pace over the next 18 months.
Fund Momentum Take
Gradient's Fund V is one of the cleaner VC stories of early 2026: a fund that was genuinely early, has the exits to prove it, and has now successfully made the structurally difficult leap from corporate venture to independent GP without losing its identity or its best people. The $220 million raise is not a stretch — it is a disciplined step up from Fund IV that keeps the firm competitive in seed rounds without forcing it to write checks that are too large for its sweet spot.
The risk here is concentration. AI application-layer investing is no longer contrarian — it is the consensus view, which means Gradient will face far more competition for the best deals than it did in 2017-2021. Every multi-stage fund now has a seed strategy, and several new dedicated AI seed funds have emerged in the last 18 months. Gradient's response to that competitive pressure is its flywheel: 500+ portfolio companies generating referrals, and a deal flow volume that most new entrants cannot match on day one.
Our read: Gradient is a buy. The independence from Google is the upgrade that should make institutional LPs more confident, not less. The exit record is real. The fund size is appropriate. And the firm's willingness to say "no" to foundational model companies and mega-seed deals is exactly the kind of discipline that separates funds with durable returns from those chasing the market. Watch the deployment pace closely over the next 12 months — if they are deploying capital well, Fund VI will be even larger.
Frequently Asked Questions
What is Gradient Ventures and how did it start?
Gradient Ventures was founded in 2017 by Google as a dedicated AI seed fund — the first of its kind at that scale. Originally operating as a corporate venture arm of Alphabet, the firm spun out as an independent entity last year, with management company ownership transferring to GPs Darian Shirazi and Zach Bratun-Glennon. Google remains an LP in Fund V alongside new institutional investors.
Who are the key LPs in Gradient's $220M fifth fund?
This is the first Gradient fund to include institutional LPs beyond Google. Alea Capital led the LP syndicate, with Reflex Capital, Smartlink, and J&T Ventures joining alongside Google, which remains invested as a continuing LP.
What does Gradient Ventures invest in?
Gradient invests exclusively in pre-seed and seed stage AI companies, specifically in applications, agentic platforms, and real-world systems that use AI as a core capability. The firm explicitly avoids funding foundational model companies and is cautious about participating in mega-seed rounds above $100 million.
What is Gradient Ventures' track record?
Gradient has backed over 500 AI companies across five funds and nearly $1.2 billion in AUM. Notable exits include CentML (acquired by NVIDIA at $400M+), Streamlit (acquired by Snowflake for $700-800M), and Prepared (acquired by Axon). Current portfolio companies include Writer, Lambda, Krea, Oura, Sona, and Airspace Intelligence.
How does Gradient differentiate from other AI-focused seed funds?
Three things set Gradient apart: its 2017 founding date gives it nearly a decade of AI-specific pattern recognition; its 500-company portfolio creates an unrivaled referral and co-investor network; and its disciplined avoidance of foundational models and mega-seed rounds keeps it focused on the market segment where it has actually demonstrated returns.
Have a fund closing to announce? Submit your fund here.
Need help raising capital? Check out our Fundraising Advisory services.