'Superintelligence' may never arrive, Deutsche Bank says

By Grace Kim

Deutsche Bank's Jim Reid says "superintelligence" may never arrive — and that AI's payoff for the U.S. economy may look very different from what markets expect.

The bank's strategist framed his outlook around a counterintuitive claim: slower, more limited AI doesn't have to be bad news. According to Reid, productivity gains from artificial-intelligence adoption have the potential to create new jobs, not just destroy them.

That's a sharp turn from the dominant narrative. Much of the current AI investment boom rests on the assumption that ever-more-capable models will keep arriving on schedule, each one closer to matching or exceeding human cognition across the board. Reid's view, as reported, questions whether that endpoint — superintelligence — ever materializes at all.

The distinction matters for anyone pricing AI exposure. If artificial intelligence plateaus as a powerful but conventional productivity tool, the economic story shifts from a race toward machine intelligence to something more familiar: a technology that spreads through workplaces, automates specific tasks and frees up labor and capital for new uses.

What does the claim actually say?

Two things, and only two, come straight from the source:

• Deutsche Bank, through strategist Jim Reid, doubts that "superintelligence" is a guaranteed destination for AI development.
• Reid sees AI-driven productivity gains as having the potential to create new jobs.

The framing is conditional. Reid isn't forecasting collapse, and he isn't calling the AI trade dead. He's repositioning the debate: even a less-than-superhuman version of AI can lift productivity, and productivity is the channel through which technology ultimately feeds wages, output and employment.

For the U.S. economy, that's the variable to watch. Productivity growth determines how fast the economy can expand without igniting inflation, and it has been the missing ingredient in much of the post-financial-crisis recovery. A technology that delivers even partial productivity gains — without reaching superintelligence — could still move that needle.

Why the jobs angle cuts against the fear trade

The employment question is where AI anxiety concentrates. Headline after headline frames the technology as a jobs killer. Reid's position, as Deutsche Bank's man on the question, runs the other way: the productivity gains from AI adoption carry the potential to create new jobs.

The mechanism is a classic one. When technology makes workers more productive, output rises. Rising output creates demand for labor in the new activities the technology makes possible. That's the historical pattern with general-purpose technologies, and it's the pattern Reid's view implies AI could follow.

Takeaway: in this framework, jobs aren't the casualty of AI adoption — they're a potential product of it.

What does this mean for the AI investment thesis?

Reid's framing doesn't invalidate the AI buildout. It recalibrates it. If superintelligence may never arrive, valuations and capital-spending plans that assume an unbounded trajectory of model capability face a real question about their terminal assumptions.

At the same time, a world of incremental AI — useful, adopted, productivity-boosting but not superhuman — is a world where returns accrue to companies that actually deploy the technology and capture the efficiency gains, rather than to those banking on a speculative endpoint.

For the U.S. economy specifically, the so-what is straightforward. Watch productivity data, not capability hype. If Reid is right that AI adoption creates jobs through productivity gains, the proof will show up in the numbers — output per hour, labor-force participation, wage growth — long before anyone settles the superintelligence debate.

His view leaves one big door open: "may never arrive" is not "will never arrive." Deutsche Bank is hedging the endpoint, not the direction of travel. The bet embedded in Reid's position is that AI's economic value doesn't depend on reaching the summit — only on keeping the productivity climb going.