Arthur Hayes, Chief Investment Officer at Maelstrom, explained on the New Era Finance podcast why Bitcoin and other cryptocurrencies are typically sold first when investors face broader market losses.
Bitcoin's Liquidity Makes It the First to Go
Hayes attributed Bitcoin's early selling to its constant liquidity and 24/7 trading. When investors face losses elsewhere and need cash quickly to cover margin calls or repay bank loans, they turn to crypto before other holdings because positions in less liquid markets cannot be closed as rapidly.
Even when investors expect future bailouts to lift prices, they require immediate cash to avoid liquidation. This selling pattern exposes Bitcoin to heavy pressure, particularly when stress concentrates in technology stocks, causing Bitcoin to track the Nasdaq closely.
The Anthropic IPO as a Market Test
Hayes described the Anthropic IPO as a key test for market confidence in artificial intelligence valuations. A strong debut could restore investor confidence and encourage risk-taking in the sector. A weak performance, however, could raise questions about sector valuations, large compute spending, and whether AI firms can generate sustainable profits.
Preparing for an AI-Capex Reversal
Hayes warned that a reversal in AI capital spending could trigger a severe global downturn. He expects governments and central banks to respond with bailouts and money creation rather than austerity measures. As a scarce asset, cryptocurrency could benefit from the resulting liquidity, though DeFi and other crypto sectors may still decline during a broad sell-off.
Until conditions improve, Hayes advises preserving cash, avoiding excessive leverage, and preparing for volatility.
Tokens for AI Compute Networks
Hayes argued that AI compute may become a commodity, with the largest long-term winners being companies built on top of the underlying infrastructure. AI agents and machines will need open, transparent, and censorship-resistant financial systems, he added.
To address this need, Hayes described FLOP, a proposed decentralized compute network where providers earn tokens for performing inference work. Agents can then spend those tokens to purchase compute services. Hayes suggested that a compute-linked token may better serve AI agents than stablecoins, which can be frozen and remain tied to traditional financial systems.
On token design more broadly, Hayes criticized projects that provide little value to users while concentrating ownership among insiders. He cited Hyperliquid's user-aligned model as a preferable approach and favors established, liquid assets and projects with proven product-market fit over speculative tokens.


