August 4, 2026
Spotify just signed Merlin -- the umbrella group representing more than 30,000 independent labels and distributors -- onto its upcoming AI remix and covers tool, joining Universal Music Group as a backer. The pitch is simple: fans get to generate AI covers and remixes of real songs, and in exchange participating artists opt in, get credited, and get paid. That last part is the whole story. For two years, AI music tools have shipped first and negotiated rights later, if at all, leaving artists to sue their way to compensation. Spotify is trying to build the licensing into the product from day one, and roping in independents alongside a major label signals it wants broad catalog coverage, not just a few marquee names.
For any business watching the AI copyright fights play out -- and there have been plenty, as we've covered in pieces like Karp's 'Marxist' Jab and the App-vs-Model Decoupling -- this is a useful template. Consent, credit, compensation, negotiated up front, isn't just good ethics; it's a liability shield. It's also a bet that fans will pay for a feature built on someone else's work if the someone else is getting cut in. Whether Spotify can make the economics work at scale, with thousands of rights holders and unpredictable usage volume, is the real open question. But the direction is right, and I'd expect other platforms sitting on user-generated remix or fan-fiction-style AI features to face pressure to match it.
Texas has been the default answer to 'where do we build the next data center' for years -- light regulation, cheap land, and a power grid operators assumed could keep absorbing demand. That assumption just broke. TechCrunch reports the state has halted new data center construction while the governor calls for audits, a striking reversal for a state that marketed itself as the anti-California of AI infrastructure. If Texas is hitting a wall, it's a signal that the entire industry's build-first strategy was resting on regional power capacity nobody had fully stress-tested.
This matters to more than utility companies. Every business relying on cloud AI inference -- which is most of them now -- is downstream of this capacity math. If states start throttling data center approvals, the cost and availability of compute could tighten in ways that ripple into API pricing and model access. It's consistent with what we flagged in AI's Money Is Flowing to Infrastructure, Not Apps: the industry's spending is concentrated in the physical layer, and physical layers have physical limits. Businesses building internal tools on top of AI vendors should treat compute scarcity as a real planning risk, not a hypothetical -- another argument for owning more of your own stack rather than depending entirely on one vendor's infrastructure roadmap, something we get into in our buy vs. build breakdown.
An analysis of seven years of Tesla earnings calls found Musk now spends roughly half his airtime on robots and AI rather than the actual car business that generates Tesla's revenue. That's not a minor tangent -- it's a reallocation of executive attention away from the product paying the bills toward the product that might, someday, justify the valuation. Investors have tolerated this because Tesla's stock has increasingly priced in Optimus and full self-driving as the real growth story, not sedans and SUVs.
I think this is worth watching skeptically rather than admiringly. There's a real difference between a CEO building genuine optionality and one using a moonshot narrative to paper over a maturing core business, and seven years of skewed call time is a long pattern to explain away as just enthusiasm. We've written before about whether Tesla's most ambitious robotics numbers hold up -- see Tesla's 10 Million Optimus Target: Real Bet or Mega-Hype? -- and this data point adds fuel to that question rather than answering it. For any executive team, the lesson isn't 'talk about AI more.' It's that credibility comes from shipping, and the market will eventually ask Tesla to reconcile the airtime with actual robot units in the field, not just projections.
Which of these three stories worries you more as a business planning around AI: the compute crunch in Texas, the rights model Spotify is testing, or a major AI-adjacent company whose leadership seems more focused on the future than the present?
Sources