Arthur Hayes: AI costs may hit middle class before midterms

Arthur Hayes on Tuesday wrote in an essay titled ‘The Butterfly Touch’ that rising costs from artificial intelligence, higher energy use and new credit creation could pressure the U.S. middle class before the November midterm elections and drive more capital into Bitcoin.

Hayes said large increases in spending on data centers, electricity, chip production, military equipment and disaster-relief infrastructure will add to inflationary pressure that voters will notice. He tied the current market dynamics to the U.S. bombing of Iran on Feb. 28 and wrote that geopolitical tension has already prompted inflows of dollars and yuan into real and digital assets.

He argued the next phase of AI deployment will require financing beyond corporate profits. Central banks and commercial lenders, he wrote, are likely to provide credit for costly computing and electrification projects. Hayes noted China has encouraged banks to shift lending away from real estate and toward technology, and he said U.S. policy also supports expanding electricity production and data center capacity.

To explain how firms will consume more computing, Hayes introduced two concepts. He called one ‘Jovan’s Paradox,’ where cheaper AI tools lead companies to use more computing power rather than cut spending. The other he labeled the ‘Red Queen Effect,’ where competitors must keep investing to match any firm that builds a superior model, quickly making prior hardware obsolete.

Hayes described political effects he expects from those economic changes. He wrote that the U.S. campaign environment could turn sharply against AI and inflation ahead of the midterms. He noted that roughly 10% of Americans own enough securities to benefit directly from gains tied to AI and tech investments, suggesting price increases and job uncertainty could become a voter concern.

On global risks, Hayes said some countries have relied heavily on dollar-denominated assets and U.S. security rather than investing in physical resilience such as pipelines, fuel corridors and food storage. He warned that if nations sell dollar assets to finance their own security, U.S. markets that depend on foreign funding for deficits could be affected.

Hayes outlined policy tools the U.S. could use in such a scenario, including dollar swap lines that let allied central banks borrow dollars and easing regulatory limits such as the enhanced Supplementary Leverage Ratio to allow banks to hold more Treasuries and equities. He placed the current stance in historical context, citing the rise of dollar reserves after the 1970s petrodollar arrangements and the Asian Financial Crisis of the late 1990s, and he said countries now favor larger buffers over minimal inventory strategies.

On Bitcoin, Hayes wrote that the cryptocurrency has outperformed gold, the Nasdaq 100, the IGV ETF and U.S. tech stocks since Feb. 28. He estimated Bitcoin had bottomed near $60,000, suggested a possible rise to $126,000, and warned the rally could accelerate past $90,000 as short positions and option sellers cover. He added he had ‘no idea how high Bitcoin can go’ and planned to take Maelstrom’s portfolio to maximum risk unless conditions change.

Hayes also noted limits to the cycle he described. He wrote the rally could end if an oversized initial public offering or a wave of mega-mergers outstrips market capacity.

All statements in the essay were presented as Hayes’s analysis of how AI spending, credit growth and geopolitical risk could affect inflation, political sentiment and asset flows ahead of the November midterms.