Jonathan Koomey warns that orbital data centers are a distraction and that AI energy demand will grow slower than expected because labs can no longer subsidize all users
Executive summary: Researcher Jonathan Koomey published an opinion piece stating that placing data centers in space is a distraction and that AI’s electricity use will rise less than projected because AI labs can no longer afford to subsidize every user. The remark calls into question the economic and energy assumptions driving major investments in both terrestrial hyperscale facilities and orbital data‑center concepts, potentially shifting capital toward efficiency improvements and prompting regulators to revisit energy‑use reporting requirements.
Who is involved: Jonathan Koomey (researcher), AI research laboratories, data‑center operators, space‑technology firms, venture capitalists, and energy‑policy regulators.
Likely next: Continued debate over AI energy forecasts, possible reallocation of investment from space‑based projects to terrestrial efficiency initiatives, and upcoming reviews of data‑center energy‑efficiency standards in the EU and US.
Koomey’s critique challenges the prevailing optimism about space‑based computing and the assumption that AI workloads will continue to explode in energy consumption. He argues that the financial model underpinning today’s AI research — where laboratories absorb costs to attract users — is reaching its limit, which will curb both the expansion of massive data centers and the rush to move them off‑planet. The statement serves as a reality check for investors and policymakers who have been allocating capital to futuristic infrastructure based on high‑growth AI forecasts.
Timeline
- — Jonathan Koomey: “La idea de llevar centros de datos al espacio es una distracción” (El País — Economía)
Analysis — what this means
Sectors affected
- Data‑center operators
- Space‑based computing firms
- AI hardware manufacturers
- Consumer‑electronics AI integration
Historical parallels
- The 2000 dot‑com boom overestimated internet traffic growth, leading to excess fiber‑optic capacity that later caused a glut.
- The 2018 AI hype triggered overinvestment in specialized AI chips that subsequently faced low utilization rates.
- The early 2010s optimism about 5G rollout speed resulted in premature network builds that were later scaled back.
Key entities
Sources
- Jonathan Koomey: “La idea de llevar centros de datos al espacio es una distracción” — El País — Economía