The AI question hiding inside a markets story
Trading venues are moving toward longer and eventually round-the-clock sessions. Brokers are adding contracts that pay out on events and company metrics. These look like market-structure stories, but they share a technical premise with the AI industry: markets that never close are traded in large part by software, and software that moves money has to prove who it is acting for. This piece follows both threads and keeps what is verified separate from what is merely announced.
The clearest hard number comes from Robinhood. In its Q3 report on October 27, event contract revenue rose more than sixfold to $173M. Over the same period crypto revenue fell 55% year on year to $120M.1 Total revenue rose 12% to a record $1.42B. Net income fell 14% to $476M, and diluted EPS fell 16% to $0.51 against $0.55 expected.1 Prediction-style products now earn more than the crypto business at this broker, which matters for any AI system built to trade, price or surface these contracts.
What the filings say about the company behind it
Robinhood's annual figures, checked against SEC filings, show how much the business changed before the event-contract surge. Revenue was $1.865B in fiscal 2023,2 $2.951B in fiscal 20242 and $4.473B in fiscal 2025,2 more than double in two years. Net income moved from a loss of $541M in 20232 to a profit of $1.411B in 20242 and $1.883B in 2025.2 Earnings per share went from -$0.61 to $1.56 to $2.05 over the same three years.2
For a reader focused on technology, the research and development line matters. Robinhood spent $349M on R&D in 2023,2 $323M in 20242 and $355M in 2025.2 Revenue more than doubled while engineering spend stayed nearly flat, a ratio that suggests the product is scaling on an existing platform. Our reading of that ratio is analysis, not a company claim.
Cash flow was less smooth. Operating cash flow was $1.181B in 2023,2 negative $157M in 20242 and $1.638B in 2025.2 Stockholders' equity was $6.696B at the end of 20232 and $7.972B at the end of 2024.2 Cash on hand has stayed in a band of roughly $4.2B to $5.8B across the quarters we hold. It was $5.829B in Q2 2023,2 $4.162B in Q2 20252 and $5.362B in Q2 2026.2
Where the sources disagree
One seam is worth showing. For Robinhood's cash at September 30, 2025, one record gives $4.331B2 and another gives $4.3B, a gap of about $31M that is rounding rather than a real dispute. We use the precise, filing-derived figure and flag the difference so readers know the lower-precision number exists.
Source reliability also varies. The Cboe and Robinhood announcement was published through NewsEOD. Across 6,520 claims we have checked from that source, 53% held up.3 The Crunchbase News pieces cited below come from a source where 49% of 41 checked claims held up.4 That is a small sample and a modest hit rate. We treat their quotes as accurately attributed statements by the named people, not as verified facts about the world.
Contracts on company metrics, and a regulator in court
On September 30, Cboe said at Robinhood's HOOD Summit in Houston that it plans to launch binary contracts tied to company-specific key performance indicators. Products tied to 23 companies are expected in October 2026, subject to regulatory approval, and Robinhood would be the first retail broker to offer them.3 Cboe's Steve Quirk said: "Earnings contracts give retail investors another tool to inform their strategies and offer an even more precise way to trade on anticipated company KPIs."3 Our dossier does not confirm that these contracts have launched, so treat them as announced and pending.
The regulatory picture is contested. On October 27 the CFTC sued Utah to block state efforts to shut down CFTC-registered contract markets under state laws. It said it has begun legal proceedings against nine other states.5 On the other side of the ledger, our records show Sequoia Capital as an investor in Kalshi, one of the venues named in this race.6 For AI builders, the practical point is that the legal basis for these products is being decided now, state by state, and any system that trades or recommends them inherits that uncertainty.
The clock: 23 hours, then perhaps 24/7
Nasdaq says the industry's move to a 23-hour trading day is expected on Sunday, December 6, 2026, pending SIP readiness and SEC rule changes. Subject to SEC approval, Nasdaq would add a 9 p.m. to 4 a.m. ET session.7 The CFTC has announced an industry roundtable on December 15 on whether derivatives markets could move to 24/7 trading.8 Longer hours leave less time for humans to supervise, and that raises the stakes for automated agents. We note this as our inference. The dossier does not say who will trade overnight.
Stablecoins are on a similar track. Coinbase set a three-day MiCA deadline, converting non-compliant stablecoins to USDC automatically and telling users not to deposit them after October 30.9 After Brazil's October 25 runoff, in which Bolsonaro, whom prediction markets expected to win, lost, Brazilian firms bought stablecoins to hedge currency risk. USDT volume on Mercado Bitcoin rose nearly 800%.10 Treasury Secretary Bessent said on October 27 that stablecoins and digital assets could reinforce dollar dominance. He said Treasury could coordinate with the SEC and CFTC on crypto rules.11
The AI layer: who is this agent acting for?
The most direct AI development in our material is Baselayer's $35M Series A, led by M13, to extend its business-verification technology to AI agents.12 Its chief executive, Jonathan Awad, describes the problem this way: "Agents spin up and they spin down. How can you trust this random one-task agent?"12 He also warns that agents unable to show they act for legitimate people or businesses will be blocked: "agents will just get blocked everywhere."12 If markets run around the clock and agents increasingly place the orders, identity checks become part of the market's infrastructure.
Awad also gives a sense of how slow this kind of business is to build. He says relationships with financial institutions typically take 12 to 18 months to establish, and large merchant partnerships can take up to 24 months.12 He says: "We've essentially streamlined 10 years' worth of selling into two years."12 That is the company's own claim, and we have not verified it.
The investor caution: do not let the model do the maths
Sean Jacobsohn of Norwest is blunt about where AI should not be used in finance: "You don't want AI doing calculations because it is not good at math. There are certain workflows it can handle where it doesn't produce precise numbers. But when you need precision, accuracy and calculations, you can't rely on AI for that."13 This is the same standard we apply here. The numbers in this piece come from filings, and no AI system computed them.
On where AI-native companies can compete, he says: "I think it would be very hard to disrupt the core products of Workday, ADP, SAP, UKG and Dayforce. But it's easier to disrupt some of their secondary products, where the category isn't their core business."13 He also sees an opening in ERP: "I think there's a potential opportunity to disrupt NetSuite and Sage. Those companies have been around for a very long time. I'm seeing more disruption downmarket, and some of these companies will eventually move upmarket."13 Finance software sold directly to CFOs may be easier to displace, in his words because "there's actually one less layer of approval when they're buying their own software, so it is a little easier to replace it when they're the direct buyer."13
Jacobsohn says he has built a Failure Museum of more than 1,500 items from failed companies and products, which he studies to understand why they failed.13 Our dossier does not cover incumbents such as Schwab launching assistants beyond the topic framing, so we make no claims about their products here.
What to watch
- December 6: whether Nasdaq's 23-hour day goes ahead, which depends on SIP readiness and SEC rule changes.7
- December 15: the CFTC roundtable on 24/7 derivatives trading.8
- Court outcomes: the Utah suit and the nine other state proceedings, which will show whether federally registered event markets can operate nationwide.5
- KPI contracts: whether the Cboe and Robinhood products receive approval and how much they add to the event-contract revenue line.3
- Agent identity: whether banks and merchants adopt verification for agents on the 12-to-24-month timelines Awad describes.12