Macro analyst and AI Macro Nexus founder Jordi Visser said the selloff in global AI and semiconductor shares from May to June 2026 should not be read as the collapse of an AI bubble. Speaking on a recent podcast, he described the move as a “speed crash” inside a structural bull market, saying it was meant to clear out crowded retail positioning and quantitative leverage rather than mark the start of a cyclical bear market.
Visser said the long-term fundamentals behind AI remain intact and that the pullback has opened up a long-horizon buying opportunity. In his framework, 2026 is still only “the first inning” for AI token demand. As consumer-facing applications improve, he expects physical hardware supply to struggle to keep pace with exponential growth in compute consumption.
Market damage looked like deleveraging, not the end of the cycle
To support that view, Visser pointed to several indicators that he said fit a textbook speed-crash pattern. Morgan Stanley’s tech momentum factor saw its 60-day volatility spike to 87. Samsung shares fell 21% in 10 days, the sharpest drop since COVID-19. He also said 87% of S&P semiconductor components were oversold on a 14-day relative strength index basis, while leveraged retail ETFs gave back 52% to 62% of their gains for the year.
He said the broader structural bull case still shows through the damage. Of the 100 AI names in his tracked portfolio, 86% were still above their 200-day moving averages. NVIDIA, he added, had recovered its 20-day moving average and printed a MACD buy signal.
For Visser, the key point is that AI infrastructure demand comes from token generation and AI agents, which he described as an exponential demand curve rather than the linear hardware demand seen in earlier boom cycles tied mainly to human usage patterns.
Memory shortages remain central to the supply story
Visser also focused on tight memory supply. He estimated Samsung’s 2026 operating profit at $217 billion, a figure he said would exceed the company’s combined total from the previous 40 years, while valuing the stock at only 11 times earnings.
He said SK Hynix’s capital raise was oversubscribed by 7x and that Micron signed $100 billion in long-term contracts. Samsung executives, according to Visser, have said memory shortages could last “beyond 2030.” He also pointed to Tesla CEO Elon Musk building factories to prepare for a memory crunch tied to an expected surge in humanoid robots before 2030.
Consumer AI agents could become the next major driver of token demand
Visser said the next phase of hardware demand will be driven by consumer AI agents. He argued that the current bottleneck in AI adoption sits more in the product and user-interface layer than in the models themselves.
Once AI agents become as naturally embedded in daily life as maps or search engines, token consumption could rise 24x, he said. As one example, he cited an AI-native company with 90 employees whose annual token spending rose from $100,000 to $11 million in just six months. He used that case to argue that enterprise adoption of blended AI model usage is nearing an inflection point.
Crypto is being repositioned as a tool of U.S. financial statecraft
Beyond AI and hardware, Visser said “programmable finance” is another core pillar in his research. He highlighted a July 2026 speech by U.S. Treasury Secretary Scott Bessent and described it as the foundation of a new economic approach, which he called the “Bessent doctrine.” In his reading, the policy direction explicitly supports stablecoins and asset tokenization, signaling that the U.S. is repositioning crypto as a powerful tool of financial statecraft.
Visser also said the passage of the CLARITY Act and progress on the GENIUS Act point to a historic turn in the regulatory environment. In that setting, he noted that BTC briefly fell below its 200-day moving average on a technical basis, yet still showed resilience and even moved higher in the face of negative news including bitcoin sales by Strategy founder Michael Saylor.
On macro policy, Visser said the Federal Reserve is very likely to raise rates in July to reinforce anti-inflation credibility, but he does not expect a large tightening cycle to follow. In his view, that backdrop is supportive for the crypto market over the longer run.
Three themes are converging at once
Visser’s broader argument is that markets are now dealing with the convergence of three forces at the same time: exponential AI demand, strict physical hardware constraints, and state-level adoption of programmable money. For investors, he said, the speed crash led by AI and semiconductor names is not only a long-term chance to reposition in technology stocks, but also a sign that traditional capital markets can no longer stay indifferent to crypto.