Bitquery counted every swap on Solana's exchanges for thirty days, through Sept. 22, 2026. The data firm flagged 58.4 percent of that volume as circular or bot-driven trading1. That flagged share equals $117.7 billion of $201.4 billion1. Real people traded the remaining $83.7 billion1. Most of the flagged total, $111.6 billion, came from wallets that bought and sold the same token inside one transaction1.
Bitquery Flags 58 Percent of Solana's Exchange Volume as Bot-Driven
Share of $201.4 billion in Solana DEX trades flagged as circular or bot-driven, Aug. 24 to Sept. 22, 2026
Bitquery calls its pattern-based count a floor estimate, built from trading patterns alone.
Source: CryptoSlate [1]. Chart by The AILately.com Desk.
The numbers behind this chart
| Item | Value |
|---|---|
| Flagged as bot-driven | $117.7B |
| Unflagged volume | $83.7B |
| Whole | $201.4B |
Read that 58.4 percent the easy way. The conclusion comes quick: Solana's exchanges now belong to bots. That reading misses who replaced the people, and who profits from the swap. Humans used to run these trades by hand. Robinhood now sells retail traders an AI agent to run the same trades inside their own account2.
Robinhood launched its AI trading accounts in May 2026. More than 150,000 customers had opened one by October2. Their agents touched Robinhood's systems nearly 30 million times a day2. Those accounts equal half a percent of Robinhood's 28.4 million funded customers5. Robinhood's "Agent Apps" also run around the clock, trading while the account holder sleeps2.
Vlad Tenev co-founded Robinhood and now runs it as chief executive. He calls the rollout a gift of professional tools. "We're making Robinhood the best place in the world for active traders by delivering tools once reserved for hedge funds, big banks, and quant firms," Tenev said at the company's September summit2.
Alex Svanevik, chief executive of the crypto analytics firm Nansen, sees the same shift coming to every exchange. "By 2028, the norm will be to trade with agents rather than clicking buttons on a user interface," Svanevik said at the Token2049 conference in Singapore3. Liz Martin, who heads Coinbase Institutional, frames agentic trading as the latest version of an old practice. "Algorithmic traders have been using machine learning forever. When I think about my old world of electronic trading, we were playing around with Nvidia GPUs to trim things out—agentic trading is the next step," Martin said3.
Jay Yu, Amy Zhao and Danning Sui ran the hardest test yet on AI trading agents in crypto. Their study for Pantera Capital, Stanford University, IC3 and Ava Labs tracked 925,323 wallets across 11 Solana trading platforms4. Inside that pool, the top 1 percent of wallets, 2,590 of them, captured 81.4 percent of all gains, worth $1.81 billion4. The remaining 99 percent split the other 18.6 percent4. Across all wallets, 62.2 percent, 575,246 of them, closed in the red4.
The Top 1 Percent of Wallets Take 81 Percent of the Gains From AI Trading Agents
Share of wallets against share of gains, 925,323 Solana trading-agent wallets studied by Pantera Capital, Stanford, IC3 and Ava Labs
Each row is one group of wallets: the blue dot is its share of all wallets, the magenta dot is its share of the gains, and the rod is the gap between the two.
Source: arXiv [4]. Chart by The AILately.com Desk.
The numbers behind this chart
| Item | Share of wallets | Share of gains |
|---|---|---|
| Top 1% of wallets | 1% | 81.4% |
| Remaining 99% of wallets | 99% | 18.6% |
Ido Ben-Natan, cofounder of the crypto security firm Blockaid, reads the same boom and expects more harm to follow. "I believe that bad guys in the world want to make money, like most people. Crypto is the easiest way to do so," Ben-Natan said at the same Singapore conference3. Agents move faster than the humans meant to watch them. Blockaid's own caseload backs the point: the firm called this its busiest year for on-chain security incidents3.
Robinhood's own terms draw the line for who answers when an agent errs. "You assume all risk for trades executed by AI agents and for any use of your data by third-party LLM providers," the company's platform terms read2. The firm keeps the upside of a product sold to everyone. Its customer keeps the downside of code built for the broadest buyer it could find. Here is Rhoden's question, asked again of a machine age: who pays, who profits, and who carries the risk.
Bitquery's number was always about more than fake trades, and it named who still places the trades once machines take the floor. Pantera's study already gives the verdict. A tiny slice of wallets wins, and most traders lose money while the platforms collect fees either way4. The Signal, AI Lately's daily email, tracks the hires, the funding rounds and the exchange rules changing shape this week. Sign up below, and watch the next shift before the headline catches up.
Sources
- Liam "Akiba" Wright, "Solana DEX Volume Spike Hides Circular Trades, and Automated Bots Are Blamed," CryptoSlate, Sept. 25, 2026, https://cryptoslate.com/solana-dex-volume-spike-hides-circular-trades-and-automated-bots-are-blamed/
- Tessa Vaughn, "Robinhood Gave AI Agents a Trading Account and a Credit Card. The Liability Gap Just Got Wider.," Forkast, Oct. 2, 2026, https://forkast.news/robinhood-gave-ai-agents-a-trading-account-and-a-credit-card-the-liability-gap-just-got-wider/
- Angelica Ang, "Norm Will Be to Trade Crypto Through Agents by 2028, Nansen CEO Predicts," Fortune, Oct. 12, 2026, https://fortune.com/2026/10/12/crypto-founders-ai-agentic-trading/
- Jay Yu, Amy Zhao and Danning Sui, "Paper Agents, Paper Gains: An Empirical Analysis of DeFi Investment Agents," arXiv, May 27, 2026, https://arxiv.org/abs/2605.29174
- Leo Sun, "Would You Let an AI Agent Trade Crypto for You? Robinhood Thinks You Will.," The Motley Fool, Oct. 6, 2026, https://www.fool.com/investing/2026/10/06/would-you-let-an-ai-agent-trade-crypto-for-you-rob/