Robinhood CEO Vlad Tenev on tokenizing stocks, expanding access to private shares, fintech's future
Summary
- Tenev says tokenization’s biggest payoff is not 24/7 trading or instant settlement, but making inaccessible, illiquid assets available. Robinhood’s France demonstration launched in 31 countries and included giveaway exposure to tokenized OpenAI and SpaceX; Tenev said Robinhood was “I think, the first” to tokenize them. The model resembles a stablecoin: hold an asset reserve, then mint and burn one-to-one tokens that can trade publicly across blockchains.
- Sacks argued that public stocks offer the cleaner regulatory starting point, while Tenev sees private shares as potentially more meaningful long term. Sacks pointed to existing disclosures and broad public ownership, as well as the GENIUS Act’s stablecoin framework signed in July, as reasons public securities are natural candidates for global, continuous blockchain trading. Private companies care who owns their shares, however, and regulators have less public information to rely on.
- Tenev frames retail ownership of private AI companies as a way to align households with technological disruption. His thought experiment: if 20% to 30% of someone’s net worth were invested in AI companies, that person would want AI to succeed rather than merely fear it. Companies worth “hundreds and hundreds of billions” currently have zero retail ownership. Citing Cathie’s presentation, he linked expected negative inflation, high GDP growth and giant productivity gains to significant labor-force disruption.
- Tenev is reluctant to control retail risk-taking. He wants accreditation relaxed toward self-certification, potentially using an explicit warning that investors could lose 100%, complete with “a skull and crossbones.” His shorthand—“No crying in the casino”—came with a firm condition: products must be clear, but opportunities available to wealthy investors should generally be available to retail.
- Robinhood’s expansion thesis is that additional financial products deepen rather than cannibalize brokerage relationships. Retirement-account users increased individual-account funding, while customers making Robinhood’s credit card top-of-wallet also moved more money onto the platform. Against a coming transfer of more than $130 trillion to younger generations, Tenev calls Robinhood’s existing quarter-trillion-plus assets “just a drop in the bucket”; the opening montage also cited 3.5 million Gold subscribers and the company’s addition to the S&P 500.
- Tenev’s separate AI company, Harmonic, targets mathematical superintelligence and formal verification. Founded two years ago, it announced International Mathematical Olympiad gold-medal-level performance and, to Tenev’s knowledge, was the only formal model to do so. Jason contrasted that result with informal models from OpenAI and Gemini; Tenev said Gemini’s informal model got silver last year. Formal methods can strengthen reinforcement-learning rewards and help verify generated software because, as Tenev put it, “human verification just doesn’t scale.”
Deep dive
1. Robinhood’s free-trading inversion has reached index scale
The opening montage set the investor backdrop: shares had more than doubled since the last report, risen 180% this year after nearly doubling in 2024, and gained more than 400% over the last year. Gold reached a record 3.5 million subscribers, and Robinhood had just joined the S&P 500.
Jason’s prelaunch memory captures the founding inversion: at a $20 million valuation, Tenev proposed attracting millennials and Gen Z to stocks while charging nothing to trade. Jason’s skeptical recap ended with “I’m in”; Tenev had told him, “This is probably the best idea you’ll ever have.”
2. Tokenization turns asset custody into a retail access layer
At “To Catch a Token” in southern France, Robinhood demonstrated an app rebuilt around crypto infrastructure: stocks on blockchains, crypto-native features including perps, and availability across 31 countries.
Tenev’s hierarchy was explicit: 24/7 trading and instant settlement have “real value,” but “the most powerful thing” is making inaccessible, illiquid assets available. Robinhood offered giveaway exposure to tokenized OpenAI and SpaceX, which Tenev said it was, “I think,” the first to tokenize.
The mechanism resembles stablecoins: keep dollars or Treasuries in a reserve, then mint and burn tokens against it one-to-one. Extending that concept to securities, the tokens can trade publicly across multiple blockchains.
Company consent remains delicate. Tenev said reactions “depend”; he had spoken with Sam before and after the OpenAI launch and believed Sam understood the goal, but conceded tokenization was a distraction from OpenAI’s mission. Robinhood is working on U.S. mechanisms while expanding in Europe, where the mechanisms will probably differ for at least some time.
3. Private ownership is the harder—and more consequential—fight
Tenev described the administration change as moving Robinhood from defense to collaboration: previously, the administration would not meet with him in person, with officials working remotely until, he thought, 2023 or early 2024. Robinhood faced one enforcement action after another and had a Wells notice; “all aspects of our business were sort of under assault.” Consumer protection was the stated rationale, while he hedged that vested interests might also matter.
Sacks’s sequencing: the GENIUS Act, signed by the president in July, supplies rules for stablecoins, while public stocks already have disclosure requirements and broad public ownership, making global 24/7 trading and instant settlement easier. Private companies care who owns their shares, and regulators have less public disclosure to rely on. Sacks’s Harbor real-estate tokenization startup, founded roughly a decade earlier, was “way too far ahead of the curve.”
Tenev said “private could be more meaningful long term.” Citing Cathie’s presentation’s projections of negative inflation, high GDP growth and giant productivity improvements, he argued that meaningful labor disruption likely follows. Placing 20% to 30% of someone’s net worth in AI companies could turn threatened outsiders into participating owners.
On responsibility, Jason pointed to just-in-time options tests and young users mixing sports betting, crypto, puts and calls. Tenev proposed self-certification and explicit 100%-loss warnings, while extending the 401(k) access order to IRAs. His bias comes from staking his career on a “maximally leveraged bet on one company.”
Jason proposed synthetic private-share exposure settling after an IPO. Tenev drew the boundary: prediction markets need an expiration date, so Robinhood currently has a market on which companies will IPO; other platforms have covered IPO prices, but direct exposure to the underlying private equity is not currently available.
4. Product convergence supports a comprehensive Robinhood platform
Robinhood’s evidence against cannibalization comes from customer behavior: opening a retirement account increased funding of individual brokerage accounts, while becoming a primary credit-card user also increased money held at Robinhood. “The two products help each other.”
That supports a comprehensive-platform ambition spanning direct deposit, investments, credit, early Gold status, family members and children—not merely replacing legacy brokerages.
Tenev says incumbent financial institutions have regulatory muscle, global scale, tens or hundreds of millions of customers and lots of assets, but can be slow to adopt new technology, lack the best engineering teams, move slowly and struggle to hire top talent. Robinhood has historically been less acquisitive, though it is doing more now, which helps it avoid multiyear integration burdens. Its test is whether it can gain incumbent scale while preserving “the nimbleness of a technology startup.”
5. Harmonic aims to make machine reasoning verifiable
Harmonic is separate from Robinhood; Tenev founded it two years ago and serves as chairman. Its target is “mathematical superintelligence”—reasoning beyond any individual human researcher—and it had announced IMO gold-medal-level performance a couple of weeks earlier. Tenev said, to his knowledge, it was the only formal model to reach that result; Jason contrasted it with informal models from OpenAI and Gemini, while Tenev said Gemini’s informal model got silver last year.
Formal verification provides a precise way to establish that a statement is true, creating a strong reinforcement-learning reward signal: incorrect data can be discarded and models trained on correct outputs. It also addresses hallucinations and software verification; when LLMs generate thousands of pages of code, particularly for back-end systems, human verification does not scale.