By Sara Flounders
October 14, 2026

While headlines obsess over artificial intelligence (AI) as an existential threat, a deeper story unfolds beneath the surface — advanced technology produces radically different outcomes depending on which society wields it. The real divide isn’t between machines and humans — it’s between profit-driven markets and strategic socialist planning that benefits billions of people rather than a few billionaires.

Infrastructure investment reveals different priorities. U.S. tech giants — Alphabet, Microsoft, Meta and Amazon — are pouring approximately $765 billion combined into AI infrastructure this year alone, with projections suggesting it could hit $1 trillion next year. According to finance.yahoo.com (June 3), these four “hyperscalers” will invest a combined $5.3 trillion from fiscal year 2025 to 2030.

Yet this investment flows almost exclusively through private channels, contingent on immediate profitability for shareholders.

China takes a fundamentally different path. The Chinese government has outlined a $295 billion commitment to build data centers over five years, complementing its broader Made in China 2025 initiative. (bloomberg.com, June 9) Since 2014, state-led semiconductor investment has exceeded $150 billion — roughly triple what the U.S. CHIPS and Science Act earmarked, according to the U.S.-China Economic and Security Review Commission.

This isn’t about quarterly returns; it’s about long-term national development.

When infrastructure decisions serve public rather than private imperatives, the beneficiaries multiply by leaps and bounds. In the U.S. capitalist system, profits concentrate among tech executives and venture capitalists. In China’s model, technological advancement functions as public infrastructure — comparable to electrification in the 20th century or internet expansion in the 1990s.

Rare earth mastery: education as strategic foundation

Rare earth magnets power everything from wind turbines to data center cooling systems. China controls roughly 60% of global mining and 90% of refining capacity. But the advantage extends deeper than resource extraction — it’s embedded in skilled human labor power development itself.

According to a Reuters investigation (May 31), at least 11 Chinese universities run dedicated rare earth degree programs, with specialized curricula covering the entire supply chain from processing to magnet manufacturing.

Keith Bradsher, Beijing bureau chief for the New York Times, reports that China also maintains 39 universities with programs to train engineers and researchers for the rare earths industry. Meanwhile, Bradsher notes, the U.S. offers no comparable undergraduate degrees anywhere.

When technology becomes embedded in education itself, it creates generational advantages that market fluctuations cannot erode. China has built an ecosystem of more than 40 specialist rare-earth laboratories producing cutting-edge research, creating a knowledge infrastructure that sustains China’s position regardless of individual company performance.

Meanwhile, the U.S. is projecting rare earth plants faster than it is building rare earth engineers.

Underwater innovation: When coordination trumps markets

Out of 32 countries that host AI data centres, around 90% of these facilities are located in China and the United States, according to a U.N. report. (tinyurl.com/e9t8t7mf)

Both China and the United States aim to be market leaders in AI and other complex computing operations. While many U.S.-based data centers are powered by fossil fuels, such as gas, China is increasingly looking to transition to green energy to meet its power needs, reduce dependence on fossil fuels and boost energy security over the coming decades. (tinyurl.com/4uwys2ep)

In June 2026, China officially inaugurated the Shanghai Lingang Underwater Data Center, the world’s first offshore, wind-powered, underwater data facility, according to oilprice.com reporting (June 20). Built with $238 million in combined investment by HighCloud Technology Team and the state-owned China Communications Construction Company, the project installs 24-megawatt server units 10 meters beneath the ocean surface. By deploying beneath seawater, the facility eliminates traditional HVAC cooling needs entirely. Connected directly to offshore wind farms just kilometers away, the center achieves near-zero fresh water consumption while delivering data to Shanghai in just 0.5 milliseconds.

Microsoft pioneered underwater data centers through Project NATIC off Scotland’s coast in 2020, testing nitrogen-filled pressure vessels at 117 feet depth. Despite successful trials, the tech giant abandoned further deployment. Why? Because underwater infrastructure doesn’t fit neatly into quarterly return models — it requires long-term coordination between energy policy, maritime zoning and computational demand that market actors rarely prioritize. (emergedge.com, May 21, 2025)

As one analyst noted in Scientific American coverage (July 2025), the difference is that a company can “springboard toward large-scale deployments of offshore, wind-powered, undersea data centers with the support of the Chinese government.”

Remote clean energy: Geographic logic versus real estate markets 

China’s data center expansion follows geographic logic rooted in resource availability, not real estate speculation. Eight designated hubs span western regions — Inner Mongolia, Qinghai, Gansu, Xinjiang — where renewable energy abundance meets sparse population density. Under the National “East-West Computing Strategy,” all new data centers in these hubs must source at least 80% of electricity from clean/green sources such as solar and wind.

The scale demonstrates coordinated planning; a pilot data center in remote Qinghai operates fully on a solar microgrid. In northern China’s desert regions, facilities connect to 500 megawatts of solar and 1.5 gigawatts of wind power.

According to Bloomberg reporting from 2022, three additional batches of desert solar and wind projects will add 455 gigawatts of capacity by decade’s end.

Contrast this with profit decisions in the U.S. In May 2026, Goldman Sachs Research forecast U.S. data center power demand jumping from 31 gigawatts in 2025 to 66 GW by 2027 — a 113% increase in just two years. Northern Virginia’s “Data Center Alley” strains local utilities, with data centers’ share of peak summer demand climbing from 4.1% to 5.3% in 2026, threatening grid stability.

Driving data center construction primarily through a short-term profit motive is leading to severe environmental and economic risks and rising anger at the staggering cost and impact to surrounding communities. The threat is real: massive energy consumption, water depletion, pollution and noise, along with tax breaks only to wealthy investors. Underneath this is the growing fear of speculative, AI-driven overbuilding of more than 5,000 projected data centers. This can leave communities with massive, underutilized infrastructure if tech demand shifts or companies abandon facilities.

Perhaps most revealing is the cost differential: China’s underwater data centers cost less than half of comparable land-based facilities, according to Inside China Business reporting. Savings come not from cheaper materials but from integrated planning, where seawater eliminates cooling costs and offshore wind avoids transmission losses. (tinyurl.com/nz6n4jw5)

The choice ahead

The question isn’t whether AI will arrive. It already has. The question is whether we accept a future where technological progress serves only those who can afford it  or whether we demand systems where innovation strengthens communities broadly.

Throughout history, technological breakthroughs have disrupted social relations built around older systems. The steam engine dismantled artisanal crafts. The automobile transformed urban geography. Each transition generated anxiety — and opportunity. AI represents no fundamental departure from this pattern.

What matters is not whether machines become intelligent, but whose interests those machines serve. In a profit-maximizing system, automation displaces workers without regard for community stability. In a socially-planned economy, technological deployment considers labor absorption, regional equity and long-term human development.

China’s infrastructure push demonstrates that technology need not follow market demands alone.

This technology is changing rapidly and globally. What is sustainable is still being tested. What is already observable is that different systems produce different outcomes.

As we stand at this technological turning point, one lesson is clear. Advanced technology always arrives embedded within social relations. The task ahead isn’t just mastering machines — it’s understanding and choosing which social relations deserve mastery. Billions of people can benefit from deliberate planning decisions rather than watching a handful of billionaires capture the rewards of collective technological advancement.