AI Sovereignty: Who Actually Wins When You Own Your Data Alone?
Alex Karp, the CEO of Palantir, went on CNBC on July 1 and called the AI industry “effing insane.” He said the CEOs he speaks with privately are livid. He said enterprises are paying for tokens that generate no value while surrendering the intellectual property that makes them competitive. He called it a wealth tax on American business.
It was a remarkable performance. And he wasn’t entirely wrong.
The day before the interview, Palantir published a nine-point AI sovereignty manifesto on X. The core argument: institutions that hand their data and model weights to third-party AI vendors are surrendering their future. Own your stack. Control your fate.
It’s a serious argument. Parts of it are right. But it was written for a specific kind of institution, one that already sits on a massive proprietary data moat. It doesn’t describe the reality of an independent specialty retailer trying to compete in 2026.
We think there’s a better model. And we’ve been building it for forty years.
Where Karp Is Right
Let’s give him the strongest version of the argument, because he’s earned it.
The weights point is the one worth taking seriously. If a vendor trains on your data and you lose access, you lose the pattern recognition you helped build. Karp frames this as allowing others to “migrate the alpha of your business to theirs.” That’s a legitimate risk, and any honest vendor, including us, should answer it directly.
The tokenmaxxing critique also lands. Per-token pricing creates a structural incentive to generate more AI output rather than better outcomes. Karp called it a wealth tax. That’s sharp language, and it describes a real misalignment that any buyer of AI services should ask about.
At M1, we answer the weights risk this way: anonymized data pooling with transparent governance. Your store’s numbers go into the aggregate without identifying your business. You can audit the methodology. You retain your own records. The intelligence that comes back to you is compounded across thousands of similar businesses, not held hostage to a subscription.
That’s a real answer to a real concern. But it only makes sense if pooling is actually possible, and for independent retail, it is.
The Flaw in the Zero-Sum Frame
Here’s where Karp’s framework breaks down for small operators.
His argument assumes you have a data moat worth protecting. It assumes that your institutional knowledge, if kept private and compounded over time, generates competitive advantage. For a Fortune 500 company, a government agency, or a defense contractor, that’s true.
A five-location women’s clothing boutique in Denver is not competing against the five-location women’s clothing boutique in Memphis. They’re not in the same market. Hoarding data from each other doesn’t create advantage. It just leaves both of them blind.
The only way a small operator gets access to the kind of pattern recognition that Walmart generates internally at scale is to pool it. And when you pool data across businesses that aren’t competing with each other, nobody loses anything.
The alpha doesn’t exist in isolation. It only exists in the aggregate.
One analyst framed the debate as “data communism” versus “data capitalism,” where data communism means everyone gains access to the same intelligence, and data capitalism means proprietary advantage stays exclusive. Karp is fighting for data capitalism on behalf of large enterprises.
For independent retail, that framing doesn’t hold. Pooling data across non-competing boutiques isn’t data communism. Nobody’s advantage is being redistributed. The advantage is being created for the first time.
Reframing Sovereignty
Karp defines sovereignty as: own your stack, control your weights, don’t depend on a third-party provider.
We’d define sovereignty differently for an independent retailer: the ability to make your own decisions, with full information, without being pressured or steered by a vendor’s incentives.
Under that definition, a retailer plugged into M1’s planning platform and Ask Indie, with access to forty years of closed, curated benchmark data, is more sovereign than one running a general AI tool trained on the open internet that has no idea what a markdown cadence looks like in specialty footwear.
Sovereignty of judgment matters more than sovereignty of infrastructure.
An independent retailer doesn’t need to own a data center. They need to own their decisions.
The Commercial Tell, Including Ours
Karp’s CNBC interview was genuine, and it aired two days after Palantir announced an expanded partnership with Nvidia to deploy open-weight sovereign AI models. The critique of token pricing conveniently sells the alternative Palantir just partnered to deliver. That’s not a reason to dismiss the argument. It is a reason to hold it to the same scrutiny you’d apply to any vendor pitch.
We’re not exempt from this dynamic. Our argument for pooled-knowledge AI is genuine, and it sells M1’s platform. The difference is whether the vendor’s incentives align with your outcomes. We believe ours do. M1’s business grows when clients perform better, because retention, referrals, and expanded planning relationships follow performance. We’d encourage you to hold us to that, the same way we’re holding Karp to it.
What This Means for Your Store
Karp is fighting the right battle. He’s just fighting it for a different army. His clients are governments, defense contractors, and Fortune 500 enterprises with genuine adversaries trying to steal their institutional knowledge. For them, sovereign infrastructure is the answer.
For an independent retailer, the answer is sovereign judgment: the ability to make better decisions than the competition, faster, using intelligence you couldn’t generate alone.
That’s what forty years of closed benchmark data, compounded across independent retailers who aren’t competing with each other, actually provides.
Sovereignty isn’t about who owns the server. It’s about who controls the decision.
Onwards and Upwards,
Marc Weiss
Co-founder and CEO
Management One
|
Pass It Along
Know another retailer who'd want this in their inbox?
We send a read like this regularly including a variety of topics indie retailers like you care about: what we're seeing across the community, what's working, what's not, and what to do about it. Forward this post, or sign up below to get it straight from us.
|