Nonrival

Expert analysis from think tanks and academia, on public policy, economics, and technology.
Human experts. AI summaries.

Data Centers Do Create Local Jobs — But Far Fewer Than Their Boosters Claim

brookings

• New research comparing counties with built data centers against those with announced-but-canceled projects finds that a first large data center typically generates only 100-200 permanent local jobs, far below what industry-sponsored reports have suggested.
• The type of facility matters: hyperscale campuses (built by Amazon, Google, Microsoft) generate broader local economic spillovers than colocation facilities, which lease space to remote tenants with little local presence.
• The findings land amid a fast-moving political backlash — including a proposed federal moratorium — driven by soaring electricity costs and strained local infrastructure, and suggest policymakers are making billion-dollar incentive decisions with little rigorous evidence.

A Flurry of AI Summits This Summer Revealed Two Competing Visions for Who Governs Global AI

brookings

• Over six weeks in summer 2026, a cascade of international AI governance events — from the G7 in France to the UN's first all-193-member AI dialogue to China's World AI Conference in Shanghai — showed a world fragmenting between US-led and China-led governance frameworks.
• China is actively building an alternative institutional architecture, with 29 countries signing on to its proposed World AI Cooperation Organization, while the US is consolidating a rival coalition through Pax Silica, now with 24 signatories and a semiconductor supply-chain credentialing pilot launching in Panama.
• Meanwhile, the Trump administration's domestic AI posture — voluntary, adoption-first, security-focused — contrasts sharply with the EU's regulatory approach and the UN scientific panel's warning that AI's software task capacity is doubling every four to seven months while reliable control methods lag behind.

Canada's AI Strategy Is Chasing the Wrong Metric — Small-Business Adoption Won't Deliver the $200 Billion It's Promising

itif

• Canada's new AI strategy aims to raise business AI adoption from 12% to 60% by 2034 and unlock $200 billion in GDP gains, but its tools — loans, grants, and advisory services — are almost entirely designed for small businesses that lack the scale to generate gains anywhere near that size.
• Productivity gains from AI depend on complementary capabilities like cloud infrastructure, data analytics, and R&D spending that most small firms simply don't have; a Statistics Canada study found no measurable productivity effect from AI once those factors were controlled for.
• The real payoff would come from deep AI adoption by Canada's largest firms in banking, rail, energy, and retail — but their barriers are regulatory uncertainty and organizational inertia, not financing, and the strategy largely ignores both.

In the News

The Supreme Court Is Likely to Decide Whether Trump Can Lock Up Millions of Long-Settled Immigrants Without a Bail Hearing

scotusblog

• Since last summer, ICE has been detaining immigrants found anywhere in the US — not just at the border — under a policy that strips them of the right to request release from an immigration judge, a dramatic break from decades of practice.
• Seven federal appeals courts have ruled the policy illegal; only two have sided with the administration, creating the kind of circuit split that almost guarantees Supreme Court review.
• The core legal question is whether a 1996 immigration law's mandatory detention provision applies only to people caught at the border or to anyone in the country without legal status — a ruling with stakes for the roughly 59,000 people ICE is detaining daily.

Analysis

U.S. Public R&D Spending Boosts Global Productivity as Much as It Does American Productivity — and Developing Countries Benefit Most

hoover

• A new study finds that U.S. government R&D spending generates roughly twice as much productivity growth globally as it does domestically, meaning America captures only about half the economic benefit of its own research investments.
• The gains are driven primarily by civilian (non-defense) R&D and are concentrated in non-OECD developing economies, which appear to leapfrog closer to the technological frontier by absorbing freely available scientific knowledge and importing capital goods.
• The findings have obvious implications for debates about cutting federal R&D budgets: the U.S. is effectively subsidizing global growth, and those spillovers are large enough to matter for the world economy.

Also Worth a Look

• What Russia's Wikipedia Clone Reveals About How Authoritarian Regimes Control Information (voxeu)
• AI's gains are large and rising, but unevenly shared (voxeu)
• Americans Are Fine With AI Helping Write Ads — But Hate It When AI Replaces Real People's Faces and Voices (gallup)
• The Public Doesn't Believe More Housing Lowers Prices — and That's a Bigger Problem Than NIMBYism (niskanen)
• Will the Open Internet Survive the War on Bots? Mapping the Debate Over 'AI Preferences' (cdt)
• Other Countries Have Figured Out Trump's Tariff Bluffs — and Are Playing Along (csis)
• Every American's Health Records Have Likely Been Stolen — and Nation-States Are Only Getting Bolder (think_global_health)