A sober look at growth quality, energy realities, and the political economy of data.

An Unsettling Question

Is artificial intelligence (AI) truly a new engine of prosperity—or a time bomb we are building together? It’s a fair question when the data show how heavily U.S. growth in 2025 leans on AI spending: figures cited on Fareed Zakaria GPS suggest AI accounts for about 40% of this year’s GDP growth, and in the latest quarter even approached 90%. Amid the euphoria, firms are racing to erect multi-gigawatt compute campuses said to be “approaching the size of Manhattan.” Is this the dawn of a productivity surge, or the first tremor of a bubble?
[1]

Framing the Problem: When Expectations Outrun Real Value

The core issue isn’t innovation itself, but the gap between financial expectations and real value. Jason Furman shows that without the surge in data-center and information-equipment investment, U.S. GDP growth in the first half of 2025 was essentially flat—an economy that looks like it’s sprinting but rests on one pillar: AI capex.
[2]
Meanwhile, McKinsey estimates global data-center spend by 2030 could approach US$6.7 trillion—the largest compute build-out in industrial history—while AI revenues remain in the “tens of billions.” That mismatch raises correction risk.
[3]

Quick Guide: AI Boom vs Bubble — At a Glance
Dimension Snapshot (2025) Why It Matters Source
GDP dependence AI credited with ~40% of U.S. growth; near 90% in latest quarter Narrow pillar of growth → vulnerable to investment swings [1]
Capex vs revenue Data-center buildout towards US$6.7T (by 2030) vs tens of billions in AI revenues Mismatched ledger → elevated correction risk if monetization lags [3]
Energy footprint Compute likened to 2–6 “Californias” of electricity demand May overwhelm grids, undercut green-transition targets [1]
Fossil backstop Memphis “Colossus” cited with up to 35 methane turbines “Green” AI leaning on fossil fuel → policy contradictions [4]
Data governance Training on public/creator data with unclear consent/compensation Risks “digital colonialism” & market-power concentration [1]
Policy brakes Clean-energy standards; stress-tests; data/antitrust; outcome-based incentives Turns hype into productivity while reducing systemic risk See Policy Ideas

A Grounded Analogy: A Giant with Feet of Clay

Imagine a magnificent cathedral rising faster than its congregation—the spire soars but the pews sit empty. Today’s AI industry resembles that cathedral: we’re building a “house of worship” for AGI, yet the collection plate—cash flow—lags the marble and gold.

Argument: Euphoria, Energy, and the Political Economy of Data

Euphoria can sideline rationality. City-scale data centers demand massive power—conservative scenarios equate usage to two “Californias,” accelerating to six. Some projects are currently propped up by fossil gas (e.g., claims of 35 methane turbines at Memphis “Colossus”), exposing a green paradox and social-license disputes.
[1]
[4]

Policy Ideas: Four Brakes to Keep Innovation Valuable

1) Clean energy & emission transparency. Minimum renewable share for large AI campuses; granular carbon reporting per compute unit; grid rules (renewables share, demand response, cooling efficiency).

2) Macro-financial stress tests. Treat mega-AI clusters like high-tech infrastructure: revenue stress testing, electricity-price sensitivity, delayed-monetization scenarios.

3) Data governance & fair competition. Clear permissions & compensation, algorithmic audits for high-impact models, and pro-innovation antitrust against over-consolidation in cloud/compute.

4) Outcome-based incentives. Public support only for AI that demonstrably cuts costs, lowers emissions, or lifts service quality in priority sectors (health, education, agriculture, manufacturing).

Indonesia: Opportunity Without the Euphoria Trap

Indonesia need not copy a “digital Manhattan.” Aim for an energy-lean, data-lean, public-value-rich AI ecosystem (frugal AI for crop prediction, logistics, social-protection anti-fraud, local-grid optimization). Allow large data centers—but tie them to renewables/efficiency and a project-linked productivity financing model. Preserve competition in cloud/compute so startups aren’t strangled by compute rents.

Conclusion: Keep Our Wits

Policy’s job isn’t to stop the music—it’s to ensure the stage is sturdy: clean power, fair data, competitive markets, sober financing. We may marvel at soaring compute towers, but what saves an economy is not the dome—it’s the foundation. Technology without wisdom is just an algorithm toward ruin. Think longer, act clearer—so today’s euphoria doesn’t become tomorrow’s crisis.

Post Views: 12,423