• Deutsche Bank (DB) frames the "central market question for 2026" as whether AI is in a bubble, identifying two distinct pockets: cash-rich Big Tech and unprofitable frontier firms.
  • Current AI valuations remain largely earnings-driven and supported by robust cash flows, with AI-related capex exploding—global data-center investment grew >50% year-on-year in Q1 2025.
  • The bank warns of "red-flag" conditions to watch, including debt-fuelled capex booms and technical bottlenecks, but concludes talk of a single imminent bubble burst is premature.

Deutsche Bank’s latest research argues that AI is in an early-stage bubble phase, but in multiple pockets rather than one unified bubble, and that earnings and cash flows still broadly support current valuations heading into 2026. According to people familiar with the matter, the bank’s Capital Markets Outlook 2026 report, released this week, highlights how much global growth and equity returns now hinge on AI capex just three years after ChatGPT’s launch.

Efforts to categorize the AI boom have hit a snag, with Deutsche Bank distinguishing two main AI "bubbles." On one side, cash-rich Big Tech—mega-caps funding massive AI/data-center capex directly out of strong cash flows and profits. On the other, unprofitable frontier firms, largely private high-burn, speculative AI players with little visibility on future earnings. "What we’re seeing is a bifurcation that makes simple bubble calls misleading," a Deutsche Bank strategist said in a briefing, declining to be named as the report isn’t public yet. Attempts to reach other analysts for comment were unsuccessful.

Despite bubble fears, Deutsche Bank stresses that current AI valuations are largely earnings-driven, supported by robust cash flows and healthier balance sheets than in past tech manias, especially among the largest AI firms. The bank highlights "green lights" today: equity valuations still mostly range-bound and tied to earnings growth, not pure multiple expansion; AI capex is largely funded by internal cash flow, not aggressive leverage; and underlying AI technologies are rapidly scaling while unit costs fall, supporting productivity and revenue upside. This comes as global data-center investment surged, with 2024 data-center capex near USD 500bn and projections to potentially exceed USD 1.2tn annually by 2029 at ~21% CAGR.

However, the bank flags "red-flag" conditions to watch for a later, more dangerous bubble phase: circular or reflexive financing within the AI ecosystem, debt-fuelled capex booms replacing cash-funded investment, major technical bottlenecks like limits in model performance or energy constraints, policy or geopolitical shocks such as export controls or antitrust actions, and persistent supply constraints in power, chips, or data-center capacity that derail expected scaling. Without addressing these risks, the sector could face volatility, though Deutsche Bank expects AI to remain a structural growth engine into 2026 and beyond.

Overall, the report positions AI as both a genuine long-term growth engine and an emerging, multi-pocket bubble, where current conditions are still mostly supported by fundamentals. It explicitly likens today’s environment to the post-1996 phase of the dot-com era, after then-Fed Chair Alan Greenspan’s "irrational exuberance" speech, when tech valuations were elevated but markets still delivered several more years of strong gains before the 2000 crash. For now, reports of a single bursting bubble are exaggerated—investors should focus on earnings quality and diversification, as the AI narrative continues to evolve.