A Star closes $450M Fund III to back AI application layer and seed-stage founders globally

A Star closes $450M Fund III

Kevin Hartz's seed firm A Star has closed its third fund at $450 million, up from roughly $300 million for each of its first two vehicles. The firm's five-person investment team will expand further, and Hartz frames the step-up in size as proportional rather than a strategy shift.

The core model stays intact: generalist, seed-stage, with checks typically running $3 million to $5 million, occasionally reaching $10 million. The firm's pitch to LPs rests on discipline rather than volume, an unusually high seed-to-Series A conversion rate and top-5% fund performance across the first two vintages. Hartz is explicit that he doesn't want to spray capital, viewing the fee-incentivized deployment model of large multi-stage funds as the thing to avoid.

“We've just announced today that we've raised our third fund. It's $450,000,000. We are sticking to our knitting. We are generalists. Certainly, in this super cycle that we're having right now, it's driven by AI and the application layer of AI, and we're investing pretty aggressively in that space. Our first two funds were about $300,000,000 each and are in top 5% kind of category.”

Thesis

AI's application layer is the dominant theme. A Star holds positions in Decagon (AI customer service) and Monaco, and Hartz describes the current environment as a super cycle driven by AI. He's direct that application-layer companies face a real threat from labs eating up the stack, invoking Microsoft's 1990s playbook of watching what developers built on Windows and then shipping a free competitor. His answer is that portfolio companies have to be the best product in the market and stay there. Decagon is his clearest example of that bet.

Sourcing

Hartz says founders are getting younger, with an increasing share still in high school when first backed. He points to Corey Levy at Z Fellows as a model for finding that cohort early. The broader sourcing imperative, getting to founders before anyone else does, becomes harder to execute as the fund and the competitive field both grow.

Incubations

A Star has developed a small number of companies from scratch, though Hartz dislikes the word "incubation." Two examples: a mortgage AI business run by Michael White, a former Block employee, and Soron, a physical security startup targeting high-end alarm and monitoring systems. The Soron bet is built on a falling-hardware-cost argument: autonomous vehicle investment is driving down LiDAR prices, making sensor-based home security commercially viable in a way it wasn't before.

Geography

Hartz ranks New York a clear second to San Francisco, citing the ability to pull Northeast university talent into a city that remains a major financial center. Austin doesn't register as a serious alternative in his view. A Star's New York portfolio includes WAP (creator services) and Antioch (physical AI and robotic infrastructure).

Services as software

On the question of AI companies with heavy services components, Hartz credits Palantir's forward-deployed engineer model, dating back to 2004-2005, as the long-run proof of concept for making that approach work. His caution is cyclical: in past boom eras, consulting-adjacent businesses grew quickly and fell hardest when markets pulled back.