$100 Billion in 90 Days: Still in the Dial-Up Days of the AI Revolution
- Rich Washburn
- Aug 2
- 3 min read
Let’s take a moment and let this sink in: in Q2 of 2025—just three months—the so-called “Magnificent Seven” spent over $100 billion on AI infrastructure. That’s not R&D. That’s not marketing. That’s just data centers.
One quarter.Three months.Roughly $1.1 billion per day.
And here’s the kicker: this is still just the beginning.
We’re in the Dial-Up Days—And It's Already Mammoth
For all the jaw-dropping numbers, for all the eye-popping scale, we’re still operating in the dial-up days of the AI era. The modem is already screeching. The connection's live. We’re online—but just barely.
This isn’t the golden age of artificial intelligence. It’s the early scaffolding. The part where we’re wiring up the towers, running fiber to the edge, laying foundations for a world that doesn’t exist yet—but soon will.
And yet look at the scope. We're building like it's 1999—only instead of DSL and copper, it's custom silicon and datacenter megacampuses. The scale of investment is beyond anything we’ve seen since the Industrial Revolution’s rail buildout. AI is not a plugin. It’s the foundation layer. And the world’s most powerful companies are betting six figures per second that it’s the future.
The Infrastructure Race Has Begun
Microsoft. Meta. Amazon. Alphabet. NVIDIA. Apple. OpenAI.
These aren’t experimental R&D shops. They’re the kings of cash flow. And when they drop $100 billion in one quarter—most of it just to expand capacity—they’re not playing.
They're racing to build the pipes, the power, and the performance needed to host the next wave of AI-native applications: models that reason, that plan, that build, that negotiate. Tools that don’t just autocomplete but orchestrate entire workflows. Platforms that don’t just enhance work but transform it entirely.
Depreciating Infrastructure, Accelerating Timelines
What makes this buildout even more dramatic is the tempo of obsolescence. These aren’t once-and-done investments. Unlike railroads or highways, GPU clusters age fast. The next wave of silicon is already in fabrication before the last one hits the racks.
So the ROI math has to be aggressive. Monetization can’t lag. The industry is betting that the revenue streams—enterprise copilots, AI-as-a-service, model licensing, agent ecosystems—will materialize just in time to justify the billions being poured into compute.
And if they don’t? This gets real expensive, real fast.
Tectonic Shifts Underway
What we’re witnessing is not a trend. It’s a tectonic shift. Entire industries are being reorganized around access to intelligence at scale.
For enterprise leaders, this isn’t about dabbling with AI anymore. It’s about rearchitecting your infrastructure, retraining your workforce, and rethinking your margins. The companies that win in this next chapter won’t be the ones with the best product—they’ll be the ones with the best intelligence loop.
For investors, the signals are loud and clear: the value chain is moving downstream to infrastructure. Compute capacity, power density, energy sourcing, and network throughput are becoming the new differentiators.
And for all of us? The way we work is already shifting under our feet. Keyboard to thoughtstream. Workflow to flowstate. Productivity to co‑intelligence.
Here’s the thing: $100 billion in 90 days isn't just a headline.
It’s a sign of the times.
It’s a flashing billboard that says, “The world is changing—fast.”
And the wild part? We haven’t even scratched the surface. The AI infrastructure race is still in its early innings. The future coming down the pipe is exponentially more complex, more powerful, and more integrated than anything we’ve built so far.
If this is what the dial-up phase looks like, just imagine what broadband feels like.
So keep your eyes open. Stay agile. And remember: this isn’t a one-time shift. It’s a full-blown transformation.
And it’s already underway.
Comments