The smart money on Wall Street has been competing on alternative data for the majority of the past ten years. receipts for credit cards. satellite images of parking lots. Outside Costco, a cellphone pings. The idea was to trade ahead of the earnings call if you could see what was going on on the ground. It worked for a while. The edge then subtly vanished as everyone began purchasing the same feeds.
It appears that BlackRock was the first to notice this. You get the impression that the company is looking inward, into its own files, rather than outward for an information advantage when you stroll through its Hudson Yards offices. The company oversees approximately eleven trillion dollars, which means it also has access to something much less glamorous but potentially more valuable: decades’ worth of internal memos, analyst notes, client calls, model revisions, and unfinished theses that never made it past a Tuesday meeting.
| Information | Details |
|---|---|
| Company | BlackRock, Inc. |
| Headquarters | 50 Hudson Yards, New York City |
| Founded | 1988 |
| Chairman & CEO | Larry Fink |
| Engineering Lead (AI) | Nish Ajitsaria |
| Assets Under Management | Roughly $11.5 trillion (2026) |
| AI Platform | Asimov — an internal research system run by hundreds of agents |
| Core Function | Autonomous monitoring of investment theses and filings |
| Industry Peers | Balyasny, Citadel, Two Sigma |
| Recent Coverage | Wall Street Journal, Business Insider, Hedgeweek |
Until recently, that archive was largely inactive. It is currently being carefully and slowly fed into an internal platform that the company refers to as Asimov. Asimov, which was developed in 2024 by equity investors collaborating with BlackRock’s AI Labs team, isn’t exactly a consumer chatbot. It is more akin to a swarm, with hundreds of agents constantly consuming research, earnings transcripts, filings, and the firm’s own confidential notes in search of anything that supports or refutes an established investment thesis. The engineering chief overseeing much of this, Nish Ajitsaria, has publicly discussed a future in which mostly autonomous systems are managed by small human squads. You can already see pieces of this powerful image on the ground.
This wager has an almost philosophical quality. By now, public LLMs have absorbed almost every word available on the internet. No one truly has an advantage if your model is identical to everyone else’s. One of the few areas that is still truly rare is private data, the kind that has been stored on a company’s servers for twenty years. According to one report, Balyasny has been moving in the same direction, asking analysts to enter their notes into a centralized portal so the AI has access to “reams of proprietary text.”
Whether any of this will yield returns large enough to warrant the investment is still up for debate. Financial AI projects typically appear fantastic in pilot and mediocre in production. Additionally, a model trained on twenty years of internal opinions inherits twenty years of internal errors, biases, and incorrect calls. This raises the issue of data hygiene. Alpha out, garbage in is not a true equation.
Even so, it’s difficult to ignore how the plot has changed as you watch this develop. Asset managers were secretly concerned a few years ago that AI would turn their profession into a commodity. BlackRock appears to be making the opposite argument now, claiming that AI, when directed inward at the appropriate archive, may be what ultimately turns scale into a benefit once more. It will likely take another complete market cycle to determine whether that turns out to be accurate or just costly.