Skip to main content

Manufacturing, utility and mining companies lose $4.2tn as AI trade reverses

16 NOVEMBER 2026·2 MIN READ·1 SOURCE

Global manufacturing, utility and mining companies have lost $4.2tn in value since June as the AI “picks and shovels” trade reversed. Companies more frequently linked in media coverage to data centre construction were hit hardest.

Manufacturing, utility and mining companies lose $4.2tn as AI trade reverses

Key takeaways · 4

  • 01

    Treat exposure to data centre construction as a source of volatility: the most frequently linked companies were hit hardest in the sell-off.

  • 02

    Compare market declines with company-level indicators: Caterpillar reported a record $78bn backlog and 75 per cent year-on-year growth in power-generation sales.

  • 03

    Do not assume that a weaker share price means existing work has disappeared: projects already started would need to be completed even if new installations faced a moratorium.

  • 04

    Account for uncertainty around policy: the effects of voter-backed US moratoriums and restrictions remained unclear.

A broad reversal in the AI trade

The companies had gained $9.4tn in value over the previous 12 months, so the sell-off erased almost half of those gains.[1] The FT analysis found that firms more frequently linked in media coverage to data centre construction were hit hardest.[1] Their AI-related products included jet engines adapted to power hyperscale data centres and fibre optics used to connect AI server racks.[1] The sector was the second-worst-performing in the S&P 500 since the summer, after technology.[1]

Sharp declines in individual companies

Caterpillar’s market capitalisation fell $214bn, or 44 per cent, from its record share-price high on June 30, according to FactSet data.[1] Nuclear power company Oklo fell 49 per cent despite signing a contract with Meta in January.[1] HVAC maker Comfort Systems and fibre-optics manufacturer Corning each lost around half their value over the period.[1] The losses reflect a reversal in market valuations; they do not, by themselves, establish that demand for each company’s products has weakened. The revenue fundamentals and potential earnings remained very strong across these companies.[1]

Backlogs complicate the sell-off

Caterpillar reported a record $78bn backlog and said power-generation sales grew 75 per cent year on year, driven by data centre demand.[1] Eaton, another electrical equipment supplier, reported that its orders had doubled year on year.[1] Several gas turbine manufacturers reported taking billions of dollars in orders during the first nine months of 2026.[1] Record backlogs meant revenue growth was already baked in until 2027.[1] Projects already under way would need to be completed even if new data centre installations faced a moratorium.[1]

Demand, policy and valuation risks

The sell-off came amid concerns that debt-financed hyperscaler spending and political opposition to US data centres could lead to a bust.[1] The effects of moratoriums and restrictions backed by US voters in the midterm elections remained unclear.[1] Tighter regulations were expected to bring more volatility, while widespread construction restrictions were considered unlikely.[1] Paul Kedrosky, a venture capitalist and MIT Institute fellow, said investors were wrong to treat traditional industrial companies like semiconductor stocks.[1] Institutional investment had piled blindly into these trades and probably fuelled overly enthusiastic valuations for some companies.[1] Joseph Dominguez called the post-election market movement an overreaction and said Democrats care deeply about protecting blue-collar data centre jobs.[1]

For professionals assessing suppliers, investments or project exposure, the sell-off can be considered alongside reported backlogs and orders: Caterpillar reported a record $78bn backlog, and Eaton said its orders had doubled year on year. Concerns about debt-financed hyperscaler spending and political opposition to US data centres, along with uncertainty about the effects of restrictions, complicate assessments of future installations and valuations.

Why it matters
For teams

AI training for your team

kju.ai is built for teams: daily sessions, team analytics, and rollout support for your organization.

How this developed

  1. 16 November 2026

    Manufacturing, utility and mining companies lose $4.2tn as AI trade reverses

Sources

AI fluency, one session a day, built for your work.