August 10, 2026
TechCrunch reported this week on Muse Glimmer, a new open-weight model from Meta that the outlet frames as a preview of Mark Zuckerberg's broader ambition: a personal superintelligence that lives with you, not just one you rent from a cloud API. The detail worth sitting with isn't the model's benchmark scores -- it's the open-weight part. Meta is drawing a line between AI you can download, inspect, and run on your own infrastructure, and AI you can only access through someone else's servers, someone else's terms of service, and someone else's uptime guarantees.
For a business owner, that distinction is not academic. Every closed model you build a workflow around is a dependency you don't control -- pricing can change, access can be throttled, the vendor can pivot. We've made this argument before about the risk of building a company on top of someone else's roadmap, and it's the same logic behind why we think about vendor fate as a real business risk, not just a hypothetical one. Open weights don't eliminate that risk, but they change who holds the leash. Whether Glimmer becomes a genuine alternative to closed frontier models or stays a research curiosity, the fact that Meta is pushing this framing publicly tells you the 'own vs. rent' debate in AI is now a boardroom question, not just a developer forum argument.
On TechCrunch's Equity podcast, historian Jill Lepore argued that Silicon Valley's leadership has misread the science fiction that shaped its worldview -- citing Elon Musk specifically -- and that this misreading is feeding into a push toward what she calls government by machines, with real consequences for democratic institutions. It's an unusually sharp critique for a tech-news outlet to run, and it's worth taking seriously precisely because it isn't coming from a competitor or a regulator with an agenda -- it's coming from someone whose job is reading texts carefully.
I don't think every AI executive quoting a novel is plotting to replace elected institutions. But Lepore's broader point lands: the people building the most consequential technology of this decade often treat speculative fiction as a blueprint rather than a warning, and that shows up in how casually some of them talk about automating judgment that used to belong to humans -- hiring decisions, content moderation, resource allocation. Businesses adopting AI internally should take the more modest version of her warning: automation without a human checkpoint isn't a shortcut, it's a liability. That's exactly why role-based access and clear approval steps matter more as AI takes on more of the work -- see our own thinking on role-based permissions as a practical, unglamorous defense against exactly the kind of unchecked automation Lepore is worried about at a civilizational scale.
TechCrunch also reported 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 country. This is the physical, unglamorous underside of the AI boom: every chatbot response and every generated image runs on electricity, and that electricity has to come from somewhere. When the somewhere is a purpose-built power plant sized for one company's compute needs, the climate math stops being abstract.
This isn't a new theme -- we've covered Amazon's power ambitions before, and the pattern is consistent: the AI industry is quietly externalizing enormous infrastructure costs onto local grids, air quality, and eventually taxpayers, while the headline product stays a clean, weightless chat window. Business leaders evaluating AI vendors rarely ask about this, but they should start. A vendor's environmental footprint is increasingly a procurement question, not just a PR one, and it's a reasonable thing to weigh alongside cost and reliability when picking a partner, the same way you'd weigh security posture or data handling before signing a contract.
What strikes me across all three stories is the same imbalance: enormous decisions -- about ownership, about governance, about the air we breathe -- are being made by a small number of companies, largely without asking the businesses and communities who'll live with the consequences. Which of these three trade-offs worries you more: who owns the model, who governs with it, or what it costs the planet to run?
Sources