Deutsche Bank on Superintelligence: When It Arrives vs What It Feels Like
Deutsche Bank Research Institute hosts research titled 'AI boom or doom: what will superintelligence feel like and when will we get there?' Deutsche Bank Research Institute
The listing identifies Jim Reid as a contributor. It names Azeem Azhar as an expert collaborator and Adrian Cox as DBRI personnel analyzing superintelligence.
That contributor detail comes from an Oct. 12, 2026 index of the institute's reports. That makes it the latest dated record among the available sources. Older coverage provides context for Reid's stated position.
On July 7, 2026, Bloomberg published 'AI Productivity Gains Are Years Away, Deutsche Bank's Reid Says'. Bloomberg Reid said artificial intelligence is likely to significantly enhance productivity, or output per hour worked, but that broader economic impact may still be years away.
Fortune reported a day later that Reid expressed optimism about AI's ability to transform the economy while cautioning that its benefits were not yet guaranteed. Fortune
The broader context here is the gap between capability and diffusion. Reid's July framing separates efficiency gains inside individual firms from aggregate productivity across the economy. Think of early electricity: a few factories wired up fast, but national output per hour took decades. One can move without the other moving.
The broader context here is how Wall Street research tries to price a general-purpose technology before it appears in national accounts, the official GDP and productivity statistics. Model benchmarks move fast. Output per hour does not. Adoption requires reorganization of workflows, capital expenditure on equipment and software, and labor reallocation. Measurement follows with a lag.
In my view, the DBRI framing is useful because it asks two different questions at once. When we get there is a forecasting problem. What it feels like is a transmission problem, meaning how AI passes through to jobs, growth and inflation. Markets tend to conflate them. Research that keeps them separate helps avoid duration errors, or betting too early or too late on how long rate moves last.
Looking at what this means for positioning, the risk is symmetric. Extrapolate early productivity anecdotes and you overstate near-term potential growth. Dismiss them because aggregates have not moved and you understate optionality, or future upside. Reid's stance leans toward the second error being less costly than the first. Optimism on direction, discipline on timing.
In my view, that discipline matters for rates and credit work. Productivity determines sustainable growth and unit labor costs, the cost of labor per unit of output. Get the timing wrong and terminal rate assumptions, or estimates of where interest rates finally settle, drift. For now, the institute's message is narrow. Superintelligence is a subject for structured inquiry, not a base case for this cycle's productivity path.
In my view, the staffing choice is deliberate in research design. Reid brings macro strategy. Cox provides in-house continuity. Azhar adds external technology expertise. That combination allows a cross-disciplinary call without leaving it inside a single silo. The test will be specification. Superintelligence has no agreed definition in economics, with possible meanings spanning automation of cognitive tasks, acceleration in total factor productivity, or a break in growth regimes. Each maps to different market variables. A useful contribution would define thresholds in advance and tie them to observable diffusion metrics rather than model scores alone.
