It was inevitable.
After watching how AI tackled tasks that once seemed reserved for brilliant humans, Instacart co-founder Apoorva Mehta decided to take things a step further. Last year, he launched Abundance — a hedge fund designed to let artificial intelligence call the shots.
Picture this: Thousands of AI bots scour the internet for trade ideas. They conduct the research, pick stocks to buy or short, size the bets, and even execute the trades.
Sure, a small team of humans builds and maintains the models, but the goal is clear: let AI run the fund. Mehta, who helped build Instacart into a household name, is betting that AI can overcome the natural limits of human investors.
As he put it, even exceptional investors “can only track so many opportunities at once, process them only so deeply, make only so many high-quality decisions.”
In theory, AI changes everything. It’s a bold experiment. Will it work?
The Promise of an AI-Driven Hedge Fund
On paper, the upside is obvious. Humans get tired, emotional, distracted. We have limited bandwidth. AI agents don’t. They can analyze thousands of data points simultaneously, spot patterns across vast datasets, and execute with cold consistency.
Quant funds have already proven that heavy automation can create enormous value — think Renaissance Technologies and others that turned systematic trading into multi-billion-dollar powerhouses. Generative AI adds a new layer: the ability to reason through complex, unstructured information like earnings calls, social sentiment, and research reports in ways that feel closer to fundamental analysis than pure number-crunching.
Mehta’s fund has reportedly outperformed multiple indexes so far, although details on the exact benchmarks he’s using remain limited. And with $100 million in seed financing and plans to eventually take outside capital, Abundance is positioning itself as an early leader in what could become a wave of AI-native hedge funds.
For public stocks, where markets are highly efficient, and oceans of data are available, this approach has real appeal. Speed, scale, and emotion-free discipline could be powerful edges.
But it’s not all smooth sailing…
The Downside
Critics, including Citadel founder Ken Griffin, have argued that generative AI isn’t yet moving the needle for hedge funds trying to beat the market. Markets are noisy, narratives shift quickly, and truly novel insights (like Griffin’s?) are rare. An AI system might excel at processing information, but it can also hallucinate, amplify biases in its training data, or struggle with black-swan events that don’t resemble past patterns.
There’s also the question of “edge.” If thousands of bots are reading the same public internet sources, how differentiated can any insights really be? And while AI can remove human whim, it can also lack the intuition, contextual judgment, and moral reasoning that seasoned investors can bring to the table during periods of extreme uncertainty.
Some strategies at Abundance already run fully on AI, while others still incorporate human involvement. That hybrid reality hints at the practical limits: full autonomy sounds exciting, but the most successful systems may still need experienced humans in the loop — at least for the foreseeable future.
Could AI Do This for Startups?
But for those of us who are focused on private markets, here’s where things get interesting...
Public stocks trade on exchanges with constant pricing, mountains of filings, analyst coverage, and real-time news. Startups? Not so much. Information tends to be fragmented and asymmetric. Often it seems almost deliberately opaque. Valuations can be subjective. Team quality, market timing, competitive moats, execution risks — all these indicators are harder to quantify.
So the natural question arises:
Could a similar army of AI agents be deployed to scour opportunities in the world of private startups? Could AI help identify the rare winners amid all the noise?
Art Versus Science
The idea is tempting. After all, AI could process far more data — deal flow, founder backgrounds, early traction signals, etc. — than any team of humans. It could run simulations, stress-test assumptions, and flag patterns from thousands of past startups.
But here’s the thing:
Private investing has always been as much art as science. The best calls often come from deep, human-led fundamental analysis — understanding a founder’s vision, assessing product-market fit in messy real-world conditions, and gauging the intangibles that spreadsheets miss.
So, next week in Part 2 of this article, that’s the tension we’ll explore.
We’ll start with the “common wisdom” you’ve probably heard: that roughly 90% of startups ultimately fail. We’ll look at what the numbers actually say, and then contrast that harsh reality with the track record we’ve built at Private Market Profits since 2016.
(Spoiler alert: Using a proven system that combines our proprietary AI-powered software with disciplined, human-powered fundamental analysis, we’ve delivered results that look very different from the grim industry averages!)
I’ll walk you through the numbers — including our actual loss rate, our number of winners, and some of the standout returns we’ve delivered — and explain how we actually pick deals.
In the meantime, I’d love to hear your thoughts. Do you believe AI will eventually run entire investment processes, end-to-end? Or will the best outcomes always come from smart humans using powerful tools?
Stay tuned for Part 2!
Happy Investing
Founder
Crowdability.com
