AI may be coming for our jobs, but not as quickly as some think… AI has had numerous hype cycles, dating back to the 1950s. The current cycle is by far the longest and largest to date in terms of its impact and capital investment. But the rate at which it's progressing is unsustainable and will eventually have to come to an end. AI infrastructure already demands massive capex for compute, energy, and data centers; yet future growth requires significantly greater investment. In 2026, the 5 major US hyperscalers will nearly 2x their AI infrastructure investment YoY to a total of $600B-$700B; data center investment is expected to exceed $4T over the next 5 years. Such capex levels can't scale indefinitely. Moreover, these investments don't directly yield scalable, sustainable economics or immediate revenue. Across major public and private AI companies, margins remain questionable, financials are often obscured by circular transactions among peers, and valuations far exceed current revenue -- tied instead to ambitious promises driven by escalating capex. Private market investors are depleting their dry powder, pushing AI firms toward greater debt financing for capex. This amplifies risks in the already fragile private credit market, and these risks spread to banks that have funded private credit funds. Meanwhile, the macro market is dealing with its own instability. Beyond overinvestment in private credit and the stock market (where M7 comprise about 1/3 of S&P 500 value), persistent inflation + stagnant wages + undersupply of housing have fueled an unaffordability crisis. Add geo-political tensions, like conflict in the Middle East (oil) and tariffs, and pressures are mounting. The AI bubble seems poised to burst amid a weakening macro landscape, likely before significant progress is made towards AGI. … So, where does this leave us? I expect the AI bubble will burst, triggering a recession before AI displaces most jobs in this cycle. While painful, it would force a much-needed market correction. The silver lining is that we buy ourselves more time until the next AI wave. More time for: - Us to use AI to supercharge our own productivity before it becomes a real threat to our job security. - Governments to put critical guardrails in place to properly regulate AI and protect us from security, data privacy, misinformation, and other vulnerabilities. - Society to prepare contingency plans for when AI does in fact come for our jobs, incomes, and livelihoods. While this isn't the most optimistic outlook, between unchecked AI dominance and a recession, the latter seems the lesser evil in the near term. Finally, from a VC perspective: AI infrastructure companies with differentiated value and defensible technical moats should survive the downturn until the next cycle begins. We've seen this resilience through past AI cycles, and we'll see it again.