Google's parent company just raised $84.75 billion in a single week — its first major equity raise in over a decade, oversubscribed so fast the company upsized the deal within two days. Berkshire Hathaway wrote a $10 billion check without blinking.

And here's the part nobody's saying out loud: raising the money was the easy part.

TLDR: Big Tech will spend over $670 billion on AI infrastructure this year, yet most of the data centers they've announced can't even break ground — because the real shortage is electricity, not cash. Google's fix is wild: it bought a solar-and-wind company and built data centers that shift their workloads to follow the available power. Here's how it works, plus a prompt that finds the hidden constraint in your own business.

A JPMorgan analysis found more than 60% of the data-center capacity planned for 2027 isn't even under construction yet — another 7% is running late. The richest companies on earth, with effectively unlimited capital, still can't build fast enough.

Why? Demand is outrunning supply. Google Cloud grew 63% year over year and is sitting on a $462 billion backlog it physically can't fill. The orders are there. The buildings aren't. And what's holding the buildings back isn't money.

It's power. A single large data center can pull as much electricity as a midsize city, and grid operators are so swamped sorting real projects from speculative ones that connection approvals have slowed to a crawl. As the CEO of Intersect — the power developer Google just bought — put it: there are racks of GPUs sitting dark because there isn't enough electricity to switch them on.

Here's where it gets genuinely clever. Instead of waiting years for the grid to deliver more electricity, Google went and got its own — then made its data centers flexible enough to follow it.

First, it bought a power company. Its $4.75 billion acquisition of solar-and-wind developer Intersect makes it the only Big Tech giant to outright own its power supplier. And the model Intersect pioneered is the real unlock: build the solar, wind, and battery storage on the same site as the data center — a "power-first" approach — so generation and compute come online together, instead of servers waiting years for a grid hookup.

Then comes the part that sounds like science fiction: Google engineers its data centers to shift computing workloads to follow the available power. When the Texas wind is blowing or the desert sun is high, the system leans into those sites; when power gets tight, it eases off or moves flexible work elsewhere. Through a new demand-response deal with Voltus, it even gets paid to dial back, freeing up as much as 100 megawatts for everyone else on the grid. The data center stops being a fixed drain and starts behaving like a tenant that politely steps aside at rush hour.

It's compute that chases the weather. And it lets Google break ground where rivals are still stuck in line.

That's why the $85 billion actually matters. Capital is only an advantage if you can deploy it — and Google is now one of the few that can turn dollars into running data centers instead of a waiting list. The money was never the bottleneck. It's the fuel. Google just made sure it owns the engine.

Here's the part that scales down to a business that isn't Google. Its breakthrough wasn't spending more — it was flexibility. Instead of demanding the world hand it more power, it redesigned its operations to flow with the power that's actually there.

Most businesses do the opposite: they brute-force their hardest constraint with more money instead of designing around it. More ad spend won't save you if you can't fulfill orders. More leads won't help if you can't deliver. And the first step to designing around a constraint is knowing what it actually is.

So we built a prompt that finds yours.

It interviews you about how your business actually runs, then surfaces the single constraint capping everything else — the one more money won't fix. Most people are surprised by what it finds.

The Bottleneck Audit →

Your docs have more contributors than ever. Engineers, PMs, support, marketing, and now AI agents. But most documentation tools force a choice: an accessible editor for the whole team, or the rigor of git-based version control for developers. That tradeoff slows everyone down.

Mintlify's editor removes the tradeoff. Writers get a visual WYSIWYG experience with slash commands and editable navigation. Developers keep their git-native workflow. Every visual edit is a clean commit, every commit appears in the editor. Changes flow both ways.

The editor also brings live collaboration and AI agents as first-class contributors:

The best documentation is written by everyone who has context. That's your whole team. And now, your agents. Try it at mintlify.com.

Score your docs!

The companies winning the AI race aren't the ones with the biggest checkbooks. They're the ones who figured out what a checkbook can't buy.

What's yours?