August 9, 2026
On TechCrunch's Equity podcast, historian Jill Lepore made an argument worth sitting with: the people building 'government by machines' aren't visionaries extending science fiction's warnings, they're bad readers who missed the point of the books they claim to love. She singled out Elon Musk as an example of someone treating cautionary tales as instruction manuals. That's a sharp cultural critique, but it's also a useful business signal. When the people setting the terms for how AI gets deployed across an economy are working from a distorted read of the genre that supposedly inspired them, the gap between stated intentions and actual product decisions widens. For a company evaluating any AI vendor's roadmap, that's a reason to judge tools by what they actually do to your data, your workflows, and your customers -- not by the founder's stated mission. It's the same instinct behind reading a vendor's security posture and trust documentation before believing the pitch deck.
TechCrunch reports that a planned Amazon data center in Texas includes an on-site power plant that could become the single largest source of climate pollution in the United States. Set aside the politics of that for a moment and look at the business logic: Amazon is choosing to build its own power generation rather than wait for the grid, because AI compute demand has outrun what utilities can deliver on a reasonable timeline. That's the real story. If the biggest, best-capitalized cloud provider in the world can't get enough power through normal channels and has to build a power plant to keep up, every company renting AI capacity downstream is exposed to the same bottleneck, just with less control over it. My take: this is a cost and reliability risk hiding inside what looks like an infrastructure story. Businesses locking into long AI contracts should be asking their vendors where the electrons are actually coming from, not just what the per-token price is.
Rippling reportedly burned through millions of dollars on AI tools in a matter of months before building an internal ROI tracker, and this week it turned that tracker into a shipped product: AI Spend Console, which monitors AI spending down to the individual employee and team level. That's a remarkable sequence -- a company sophisticated enough to run HR, IT, and finance software for thousands of businesses still got blindsided by its own AI bill. I think that's the headline, not the product launch. If Rippling needed a dedicated console to see where its AI money was going, most companies without that kind of internal tooling are almost certainly flying blind right now. This is exactly the kind of visibility gap that belongs in the same conversation as analytics and reporting and HR and people ops -- AI spend is now a line item that needs the same governance as headcount or software licensing, not a rounding error nobody tracks until it's too late.
Where do you land on this: is it smarter to build in-house visibility into your team's AI usage now, or wait and buy a dedicated tool once your spend actually gets painful?
Sources