In partnership with

Find out why 1M+ professionals read Superhuman AI daily.

In 2 years you will be working for AI

Or an AI will be working for you

Here's how you can future-proof yourself:

1. Join the Superhuman AI newsletter – read by 1M+ people at top companies

2. Master AI tools, tutorials, and news in just 3 minutes a day

3. Become 10X more productive using AI

Join 1,000,000+ pros at companies like Google, Meta, and Amazon that are using AI to get ahead.

Sign up and start learning AI

The Power Crisis That Could End America's AI Dreams

Your smartphone buzzes with another AI-powered notification. Your smart home adjusts the temperature automatically. Your car's navigation system reroutes you through traffic using machine learning. But what if I told you that every single one of these conveniences—and the entire future of American technological leadership—hangs by a thread so fragile that one expert described it as the difference between "hitting grand slams" and barely getting "on base"?

The thread is power. And it's about to snap.

The $6.7 Trillion Reckoning

Here's what keeps AI executives awake at night: By 2030, artificial intelligence will consume as much electricity as 22% of all American households combined. That's not a projection—that's a mathematical certainty based on current growth trajectories.

And here's the kicker: We don't have it.

While you've been watching the AI revolution unfold through ChatGPT queries and viral TikToks, a quiet crisis has been building in the backbone of modern civilization—the electrical grid that powers everything. A single ChatGPT query uses 10 times more energy than a standard Google search. Multiply that by billions of queries, add the massive data centers training the next generation of AI models, and you get a demand curve that looks less like gentle growth and more like a vertical cliff.

McKinsey's latest analysis is brutal in its clarity: Companies worldwide need to invest $6.7 trillion in new data center capacity between 2025 and 2030. That's not billion—that's trillion, with a 'T.' For perspective, that's roughly the entire GDP of Germany and Japan combined, and it needs to happen in just five years.

But money alone won't solve this problem. Because while America scrambles to find the cash, China has already built the infrastructure.

The question isn't whether America can afford to win the AI race—it's whether we can afford to lose it...

The Devastating Truth from Behind Enemy Lines

David Fishman didn't expect to return from China with a warning that would shake the foundations of American tech policy. A longtime energy expert who has spent years analyzing both superpowers' electrical systems, Fishman came back from his latest research trip with a message that no one wanted to hear but everyone needed to understand.

"China is set up to hit grand slams. The U.S., at best, can get on base."

Those weren't the words of a pessimist or a foreign propagandist. They came from someone who has made a career of understanding energy infrastructure, and what he discovered in China left him "stunned."

Here's what Fishman found: China operates with an unprecedented 80-100% nationwide reserve margin. Translation? They have at least double the electricity capacity they currently need, sitting idle, waiting for demand. When a new AI data center wants to plug into the grid, China doesn't see a problem—they see a solution. These massive computational facilities help "soak up oversupply" in a system that was deliberately overbuilt for exactly this moment.

Meanwhile, back in America, regional grids typically operate with only a 15% reserve margin. During extreme weather—which, let's be honest, is becoming the norm rather than the exception—that margin can disappear entirely. States like California and Texas regularly issue warnings about potential blackouts when demand rises even modestly.

Now imagine trying to plug in a facility that consumes as much electricity as a small city.

But China's advantage goes deeper than just having extra capacity. They've fundamentally reimagined how a 21st-century power grid should work...

The Infrastructure Revolution You Never Heard About

While America debated the merits of various infrastructure bills, China was quietly building the electrical equivalent of the Interstate Highway System—except faster, smarter, and more comprehensive than anything the world has ever seen.

The numbers are staggering:

In 2024 alone, China added 278 gigawatts of solar and 46 gigawatts of wind capacity. To put that in perspective, the entire United States has about 1,200 gigawatts of total electricity capacity from all sources combined. China added nearly a quarter of America's total capacity in a single year, and they did it entirely from renewable sources.

But here's where it gets truly mind-bending: In May 2025 alone—just one month—China added 93 gigawatts of solar capacity. That's nearly 100 solar panels installed every second, 24 hours a day, for an entire month. They added another 26 gigawatts of wind power in the same 30-day period.

China now operates 1.4 terawatts of solar and wind capacity, representing 44% of all utility-scale renewable capacity on Earth. They're not just leading the renewable energy revolution—they are the revolution.

And they've built the transmission infrastructure to match. China has constructed 42 ultra-high voltage transmission projects that can move electricity across the entire country with losses of just 4.37%—better than America's 5% average. These "bullet trains for power" can deliver electricity from solar farms in the desert to AI data centers on the coast in just three milliseconds.

Think about what this means: China can generate power wherever it's cheapest and most efficient, then move it instantly to wherever demand appears. They've created a true national grid that treats electricity like water flowing through pipes, finding the most efficient path from generation to consumption.

America, meanwhile, is still trying to figure out how to connect states that don't want to talk to each other...

The American Grid: A 60-Year-Old System in a 21st-Century Crisis

Your grandmother's house and your smartphone use the same electrical grid. Let that sink in.

Much of America's electric grid was built in the 1960s and 1970s and is now approaching the end of its 50-80 year lifecycle. We're trying to power artificial intelligence with infrastructure that was designed when humans were still using typewriters and carbon paper.

The statistics are sobering: The average American experiences about four hours of power outages annually, with the frequency of significant outages doubling over the past 20 years. And this is during normal times, with normal demand, from normal sources.

Now add AI to the equation.

AI data centers currently consume approximately 4.4% of total U.S. electricity, projected to increase to 6.7-12% by 2028. In some regions, data centers are already creating bottlenecks that would have been unimaginable just five years ago. In Virginia, residents could face $14-37 monthly bill increases by 2040 due to data center demand. In Columbus, Ohio, households are looking at an additional $20 monthly ($240 annually) just to subsidize Big Tech's energy needs.

And here's the part that should terrify anyone concerned about American competitiveness: There's currently a seven-year wait for some grid connection requests.

Seven years. In the technology world, seven years is multiple generations of innovation. By the time a data center that applies for grid connection today actually gets powered up, the AI models it was designed to run will be as obsolete as a flip phone.

But the real crisis isn't just about capacity—it's about what happens when an entire economy starts subsidizing the dreams of a few tech giants...

The Great Subsidization: Why Your Electric Bill Is Funding Silicon Valley's AI Dreams

Here's a dirty secret that utility companies don't want you to know: Data centers in 16 states receive specialized rates that aren't nearly enough to cover the cost of new power plants needed to serve them. The shortfall doesn't just disappear—it gets passed on to residential and commercial customers who have no choice but to pay.

You're subsidizing Amazon's AI experiments. You're paying for Google's ChatGPT competitor. You're funding Microsoft's push into artificial general intelligence, whether you want to or not.

Electricity costs rose 6.5% nationally between April 2024 and 2025. That might not sound like much, but for a typical household spending $150 monthly on electricity, that's an extra $117 per year. For millions of families already struggling with inflation, it's the difference between financial stability and financial stress.

Data centers accounted for 70% ($9.3 billion) of increased electricity costs in the mid-Atlantic grid last year. Think about that: Seven out of every ten dollars in increased electricity costs in one of America's most populated regions went to support facilities that directly benefit a handful of tech companies and their investors.

Meanwhile, those same companies are making record profits from AI services while everyday Americans pick up the tab for the infrastructure that makes those profits possible.

And the worst part? We're just getting started...

The Corporate Exodus: When Big Tech Gives Up on the Grid

When the titans of technology start building their own power plants, you know the existing system is broken beyond repair.

Microsoft committed to restarting the Three Mile Island nuclear plant with a 20-year power deal. Yes, that Three Mile Island—the site of America's worst nuclear accident. The fact that Microsoft is willing to resurrect a facility with that history tells you everything about how desperate the situation has become.

Amazon announced plans to deploy 5 gigawatts of small modular reactors by 2040. For context, that's enough power for roughly 4 million homes. Amazon isn't just thinking about supplementing grid power—they're planning to become a significant electricity producer in their own right.

Google signed the world's first agreement to buy power from small modular nuclear reactors. The company that once prided itself on "organizing the world's information" is now in the nuclear power business.

But perhaps most telling is Elon Musk's approach with xAI. The company imported an entire power plant from overseas and installed 35 gas turbines producing 420 megawatts to power their existing facility. They're already planning facilities requiring up to 2 gigawatts—enough to power 1.5 million homes.

When America's most successful companies start importing power plants from abroad to avoid dealing with the U.S. grid, that's not entrepreneurship—that's an indictment of national infrastructure policy.

These corporate solutions might work for billion-dollar companies, but what about everyone else?

Three Scenarios for America's AI Future

Where does this crisis lead? Based on current trends and expert analysis, here are three increasingly likely scenarios for how America's power grid crisis will reshape the AI landscape:

Scenario 1: The Great Concentration (Most Likely)

By 2030, AI development becomes concentrated in a handful of locations where companies have managed to secure private power sources. Microsoft runs its most advanced models from facilities powered by refurbished nuclear plants. Amazon's AI services run primarily from company-owned renewable installations. Smaller companies and startups find themselves locked out of AI development entirely, unable to access the massive computational resources that require dedicated power infrastructure.

Innovation slows as the AI sector consolidates around companies wealthy enough to build their own utilities. Regional economies outside these power-rich zones struggle to participate in the AI revolution. America maintains technological leadership, but at the cost of broad-based innovation and economic opportunity.

Scenario 2: The Infrastructure Reckoning (Possible)

A major grid failure during peak summer demand—perhaps in Texas or California—creates a political crisis that finally forces comprehensive infrastructure reform. Congress authorizes the $720 billion in grid spending that Goldman Sachs estimates is needed by 2030, but the mobilization takes years to have meaningful impact.

Meanwhile, China continues building capacity faster than America can plan projects. By the time U.S. infrastructure catches up, Chinese AI companies have established insurmountable advantages in computational resources and energy costs. America becomes a technology follower rather than leader, licensing innovations developed in Beijing and Shenzhen.

Scenario 3: The Collapse (Low Probability, High Impact)

The gap between AI energy demand and grid capacity becomes so severe that it triggers broader economic disruption. Rolling blackouts become common in major metropolitan areas as utilities choose between keeping the lights on for residents or powering data centers. Public anger over subsidizing tech companies while enduring service disruptions leads to political backlash against AI development.

America imposes severe restrictions on data center construction and AI model training, effectively ceding technological leadership to countries with better energy infrastructure. The economy suffers as AI-powered productivity gains flow primarily to foreign competitors.

The choice between these scenarios isn't inevitable—but it requires understanding what we're really competing against...

China's Strategic Masterstroke: Building the Future Before It Arrives

Here's what American policymakers still don't understand about China's approach to the AI race: They're not trying to build today's energy needs—they're building tomorrow's.

China's approach to infrastructure development follows a fundamentally different philosophy than America's market-driven model. They plan long-term and build in anticipation of demand, with renewable energy framed as a cornerstone of economic strategy rather than moral imperatives.

State ownership of most power plants enables coordinated planning and strategic overcapacity. When AI demand surges in a particular region, China can easily reactivate idle coal plants or redirect new solar capacity. Their energy system is designed for agility and expansion, not optimization around current demand.

This isn't just about having more electricity—it's about having the right kind of electricity infrastructure for an AI-powered economy. China has built a grid that treats electricity like a commodity that can be produced anywhere and consumed anywhere else, instantly, with minimal loss.

America, by contrast, has built a grid that reflects our political and economic system: fragmented, locally controlled, optimized for steady-state operation rather than rapid growth or change.

The U.S. political and economic system is not structured to build future grid infrastructure without public financing to de-risk long-term investments. Private companies won't invest in capacity that won't be needed for years, and public utilities lack the capital for massive speculative building projects.

Permitting delays, local resistance, and short-term investor expectations create major bottlenecks that China simply doesn't face. While China can approve and begin construction on a new transmission line in months, similar projects in America can take a decade or more.

The result is a strategic mismatch that goes to the heart of how different societies organize economic development...

The Hidden Cost: What AI Supremacy Really Means for Your Life

Strip away the rhetoric about technological leadership and global competition, and you're left with a fundamental question: What does it mean for ordinary Americans if China wins the AI race because they built better electrical infrastructure?

The implications go far beyond national pride or tech industry profits.

Data center spending is already displacing consumer spending in terms of GDP impact. This should concern everyone because consumer spending typically represents two-thirds of the U.S. economy. When massive capital investments flow toward AI infrastructure instead of the products and services that regular people buy, it represents a fundamental shift in how the economy operates.

Consider what happens when Chinese companies dominate AI development because they have cheaper, more reliable electricity:

• Your job might depend on AI tools developed in Beijing rather than Silicon Valley

• American companies may need to license AI capabilities from Chinese competitors

• The data that powers AI models—including your personal information—might flow through servers controlled by foreign entities

• The economic benefits of AI productivity improvements accrue primarily to other countries

David Fishman's warning wasn't just about electricity: "U.S. policymakers should hope that China remains a competitor rather than an aggressor, because right now they cannot effectively compete on energy infrastructure."

That's the real stakes of the power grid crisis. It's not just about whether American companies can run AI models—it's about whether America can maintain economic and technological sovereignty in an AI-powered world.

But there's still time to change course, if we're willing to think differently about infrastructure, investment, and national priorities...

The Path Forward: Lessons from Interstate Highways and Moon Shots

America has solved infrastructure challenges of this magnitude before. The Interstate Highway System transformed the economy and society in ways that weren't fully appreciated until decades after construction began. The Manhattan Project and the Apollo program proved that American can mobilize resources and expertise for strategic technological goals that private markets alone wouldn't pursue.

The power grid crisis requires a similar approach—but with 21st-century understanding of how to blend public coordination with private innovation.

Here's what a serious response might look like:

Federal Grid Authority: Just as we don't let states build their own interstate highways, America needs coordinated national leadership for electricity infrastructure. There's currently no single entity organizing the U.S. grid, unlike federal oversight of the Interstate Highway System. This fragmentation is killing our competitiveness.

Strategic Overcapacity: Instead of building just enough capacity to meet projected demand, America should adopt China's approach of deliberate overbuilding. One-fifth of U.S. counties now restrict renewable energy development—a luxury we can no longer afford when national competitiveness is at stake.

Accelerated Permitting: Environmental review and local consultation remain important, but they can't take seven years when technological leadership hangs in the balance. Emergency streamlining procedures for strategic infrastructure projects could compress approval timelines to months rather than years.

Public-Private Partnerships: Rather than expecting private companies to build speculative capacity or forcing taxpayers to subsidize tech giants, innovative financing structures could align public strategic interests with private efficiency and capital.

The technical solutions exist. The question is whether America has the political will to implement them before it's too late.

The Choice That Can't Be Unmade

As you read this, somewhere in China, another solar panel is being installed. Another transmission line is being energized. Another data center is coming online with abundant, cheap electricity.

And somewhere in America, another AI startup is discovering that their brilliant algorithms are worthless because they can't get the power to run them.

The power grid crisis represents more than a technical challenge—it reflects fundamental differences in how societies organize themselves for long-term competition. China spent decades building energy infrastructure in anticipation of future demand. America optimized for efficiency and market-driven development.

Both approaches have their merits, but only one is winning the race to power artificial intelligence.

The question isn't whether America can eventually solve its power crisis—the question is whether we can solve it fast enough to maintain technological leadership in the most important innovation race of the 21st century.

Every month of delay makes China's advantage more insurmountable. Every new data center that can't find power represents missed opportunities for American innovation. Every household that pays higher electricity bills to subsidize insufficient infrastructure represents a failure of strategic planning.

The race for AI supremacy may ultimately be won not by superior algorithms or computing power, but by whichever nation can most effectively power the massive data centers that these technologies require.

Right now, that nation isn't America.

But it could be, if we're willing to think as strategically about electricity as we once thought about highways, space exploration, and nuclear technology. The choice is ours—but we need to make it now, while there's still time to change the trajectory.

The future of American technological leadership isn't being decided in Silicon Valley boardrooms or university laboratories. It's being decided by the fundamental question of whether we can keep the lights on.

And right now, we're losing.