BMO's Jennifer Lee Calls the AI Slowdown a Pause, Not a Reversal

BMO Capital Markets economist Jennifer Lee said on Bloomberg This Weekend that calls to slow AI development may create near-term market uncertainty but that the technology's infrastructure, investment and demand are unlikely to reverse course, with labor shortages and automation keeping the longer-term expansion intact.

The argument over whether artificial intelligence should be developed more slowly has moved from conference panels into the pricing of risk assets. BMO Capital Markets economist Jennifer Lee, speaking on Bloomberg This Weekend, offered a distinction that matters for anyone holding the megacap technology complex: the pace of AI growth is negotiable, but the direction is not.

Lee said calls to slow AI development may create near-term uncertainty for markets, while the technology's infrastructure, investment and demand are unlikely to reverse course. Concerns about jobs, data centers and profitability, in her framing, act as a brake rather than a steering wheel. Labor shortages and the push toward automation, she said, keep AI's longer-term economic expansion intact.

A brake on pace, not a change of direction

That is a more specific claim than the usual bull-versus-bear framing, and it is worth unpacking. A slowdown in the rate of AI deployment shows up in earnings as deferred revenue recognition, longer sales cycles and a slower ramp in utilization of already-built capacity. A reversal would show up as cancelled data-center leases, written-down equipment and capital spending guidance cut rather than pushed out. Those are different events with very different consequences for shareholders, and markets do not always price them separately.

Lee's three named pressure points invite three separate tests. Jobs: political and regulatory pushback intensifies if displacement becomes visible in aggregate labor data rather than in individual company announcements. Data centers: the constraint is increasingly physical interconnection queues, power availability, water, local permitting rather than financial. Profitability: the question that has hung over the sector since inference costs became a line item is whether the revenue generated by AI services covers the depreciation on the hardware that produces them.

None of those three, on Lee's reading, is fatal. Each is capable of stretching the timeline. Investors who own the trade at current multiples are effectively underwriting a timeline, which is why the distinction is not academic.

Where the tape stood going into the debate

The market did not go into this conversation in a defensive crouch. At the last close before Lee's remarks, on Friday, September 11, 2026 at 20:00 GMT, the S&P 500 tracker (NYSEARCA: SPY) finished at $764.29, up 0.85% on the day from a previous close of $757.83, having traded between $763.60 and $766.38. The Nasdaq 100 tracker (NASDAQ: QQQ) closed at $714.88, up 0.87% from $708.69, with a session range of $713.63 to $717.63. The Dow tracker (NYSEARCA: DIA) closed at $525.79, up 0.97% from $520.75.

Two features of that snapshot are worth noting. First, the tech-heavy Nasdaq 100 proxy and the broad market proxy gained almost identically 0.87% against 0.85% which is not the pattern of a session driven by a scare specific to AI names. Second, the Dow proxy led all three. When the most concentrated AI exposure is not the day's outperformer and the most industrially weighted index is, the market is not expressing acute anxiety about a slowdown in AI capital spending; it is expressing something broader.

Day ranges were also tight across all three benchmarks relative to their closing levels, and each closed near the middle-to-lower part of its intraday band. That is the signature of a market drifting rather than repricing.

Why the labor-shortage argument does the heavy lifting

The most load-bearing part of Lee's case is the one that has nothing to do with model capability. If employers face structural difficulty filling roles, the business case for automation does not depend on AI systems getting dramatically better it depends only on them being adequate and cheaper than the alternative, which is an unfilled position or an expensive one.

That framing decouples AI demand from the model-release cycle. It means a deliberate slowdown in frontier development the thing the "pause" advocates are actually asking for would not necessarily reduce enterprise adoption of the tools that already exist. Deployment and frontier research are separate spending lines with separate justifications, and conflating them is one of the more common errors in this debate.

The counterweight, which Lee named, is jobs. Automation driven by labor shortage is politically defensible in a way that automation driven by headcount reduction is not. Should the mix shift from the former to the latter in visible ways, the regulatory pressure that today looks like a rhetorical campaign becomes something with legislative teeth. That is the channel through which "tempering the pace" could become sharper than investors currently expect.

What would actually falsify the thesis

Investors do not have to take the argument on faith. A handful of observable items would distinguish a pause from a turn.

• Capital spending guidance. Watch whether hyperscalers push planned spending into later periods or cut the total. Deferral supports Lee; reduction does not.
• Data-center commitments. New lease signings, power purchase agreements and site announcements are the physical evidence. A slowdown in new starts while existing builds continue is consistent with tempering.
• Depreciation and useful-life assumptions. Any shortening of assumed equipment life, or a write-down of installed capacity, would be a genuine reversal signal.
• Labor data. If measurable displacement appears in the aggregate numbers, the political constraint Lee identified moves from theory to practice.
• Index leadership. A sustained gap opening between the Nasdaq 100 proxy and the broad market would mark the point at which the market stops treating this as a sector-wide timing question.

The uncertainty is the point

Lee's own qualifier is the part most likely to be proved right soonest: the near term is uncertain. Debate about slowing AI development does not have to succeed to move prices. It only has to make the timing of returns harder to model, and that alone widens the range of plausible valuations for companies whose current prices embed a specific ramp.

The full conversation aired on Bloomberg Technology. What Lee is describing is a market that has to hold two ideas at once: that the multi-year build-out is real and financed, and that the path there will be interrupted by arguments about jobs, power and whether the economics work at scale. Those interruptions are where the volatility lives. The distinction between a slower climb and a descent is the one that determines whether volatility is an entry point or a warning.

Key facts

• Nasdaq 100 (QQQ) last close: $714.88, +0.87%, as of Fri 11 Sep 2026 20:00 GMT
• S&P 500 (SPY) last close: $764.29, +0.85%, prev close $757.83
• Dow 30 (DIA) last close: $525.79, +0.97%, prev close $520.75
• Source of the call: BMO Capital Markets economist Jennifer Lee, Bloomberg This Weekend

Frequently asked questions

What did Jennifer Lee actually say about the AI slowdown?

Speaking on Bloomberg This Weekend, the BMO Capital Markets economist said calls to slow AI development may create near-term uncertainty for markets, but that the technology's infrastructure, investment and demand are unlikely to reverse course. She added that labor shortages and the push toward automation keep AI's longer-term economic expansion intact.

What is the difference between AI growth slowing and AI investment reversing?

A slowdown means spending and deployment are stretched over a longer timeline: deferred capital expenditure, longer enterprise sales cycles, slower ramp of existing capacity. A reversal means cancelled projects, written-down equipment and cut budgets rather than delayed ones. The two look similar in headlines but have very different consequences for company earnings and valuations.

Which concerns did Lee identify as capable of tempering AI's pace?

She named three: jobs, data centers and profitability. Employment displacement raises political and regulatory pressure; data centers face physical constraints on power, land and permitting; and profitability turns on whether AI service revenue covers the depreciation of the hardware generating it. Lee framed these as brakes on pace rather than reasons for reversal.

How were the major US benchmarks trading before her comments?

At the last close, Friday September 11, 2026 at 20:00 GMT, the S&P 500 tracker SPY finished at $764.29, up 0.85%. The Nasdaq 100 tracker QQQ closed at $714.88, up 0.87%. The Dow tracker DIA closed at $525.79, up 0.97%, leading the three. Markets were closed at the time of writing.

Why does the labor-shortage argument matter to the AI investment case?

If employers struggle to fill roles, the business case for automation does not require AI models to improve dramatically. It only requires existing tools to be adequate and cheaper than an unfilled or expensive position. That decouples enterprise AI demand from the frontier model-release cycle, meaning a research slowdown would not automatically reduce deployment spending.

What should investors watch to test whether the thesis holds?

Key indicators include whether hyperscalers defer capital spending or cut it outright, the pace of new data-center lease and power agreements, any shortening of assumed equipment useful life or capacity write-downs, aggregate labor data showing measurable displacement, and whether a sustained performance gap opens between tech-heavy and broad market indices.

Sources

• Investors See AI Slowdown as a Temporary Step — Bloomberg Technology

Photo: Jan van der Wolf · Pexels Licence — source