A small circle of the world’s richest technologists has started speaking about war, deterrence, and global order in the same register they use for product launches: as engineering problems with technical solutions, where the state is a slow, sentimental client and the software company is the one actually keeping the peace. Elon Musk, Palantir’s Alex Karp, and Google’s former chief Eric Schmidt do not agree on much, and they are not a coordinated bloc. But laid side by side, their interviews, manifestos, and conference remarks describe a consistent worldview, and, just as consistently, a business model: cast state institutions as too slow to manage existential risk, then sell the acceleration that only a private technology firm can supposedly provide.
Musk’s Techno-Fatalism
Musk’s version of the worldview surfaced most vividly in a combative sit-down with The Economist’s editor-in-chief, Zanny Minton Beddoes, published this week and broken down by Gizmodo. Pressed on the future of artificial intelligence, Musk insisted AI will exceed human intelligence within five years and that humans will no longer be in charge of the world within ten, describing a coming “quasi-infinite economy” of robots and digital intelligence so vast that money itself would become largely meaningless. It is a framing in which political authority, economic policy, and democratic accountability are treated as transitional arrangements soon to be superseded by sheer computational abundance.
That same interview pushed Musk’s technological determinism into explicitly civilizational territory. Musk maintained, in an exchange later dissected by the Irish Star, that civil war in Britain was inevitable given the country’s demographic trajectory, dismissed Beddoes’s home-country rebuttal by telling her she lived a “cloistered existence,” and argued Britain needed a dissolution of parliament and a fresh election because the country could not wait years for the next one. Existing governments, in this telling, are too slow and too compromised to act on problems Musk considers self-evident, leaving unaccountable technologists as the only ones willing to say so out loud.
Karp’s Deterrence Pitch
Karp arrives at a similar diagnosis of institutional failure from a different angle, built around the claim that deterrence itself needs a new engine, supplied by his own industry. In the opening essay of Karp and co-author Nicholas Zamiska’s book “The Technological Republic,” excerpted by The Free Press, the authors invoke Alfred Nobel’s hope that dynamite would make war too terrible to fight, then argue that America’s commitment to military supremacy, and to deterrence as a legitimate doctrine, has become dangerously unfashionable in the West. Their prescription is a new Manhattan Project: a formal fusion of Silicon Valley and the Pentagon, on the premise that only private software firms can now build the weapons capable of keeping the peace, conveniently with Palantir first in line to build them.
Karp has pushed the same logic further in more recent remarks, framing the choice facing Washington in moral rather than merely technical terms. In comments reported by Fortune, Karp argued the U.S. is likely headed toward simultaneous confrontation with China, Russia, and Iran, and that nuclear deterrence has become less effective because the West is unlikely to use such weapons while adversaries might, leaving what he called a moral disparity that outweighs any technological parity. The conclusion he draws is that restraint has become a liability and that autonomous weapons, built at speed by companies like his own, are what closes the gap.
Sovereignty as Sales Pitch
Palantir’s institutional voice has recently shifted from martial rhetoric toward a subtler argument about sovereignty, though the underlying interest is the same. In a nine-point statement posted to X and backed by a dedicated page on Palantir’s own site, the company argues that an institution’s control over its data and AI model weights determines its future, warning that “data retention is your treasure” and that outside vendors will exploit any control an institution hands over. It coins the term tokenmaxxing to describe what it calls an addictive, false sense of progress from renting compute by the token rather than owning infrastructure outright, and warns against what it calls technopoliticization, in which political preference rather than technical judgment drives infrastructure decisions, arguing that only institutions with a proven track record, not popularity, should be trusted to make that call. The irony sits close to the surface: a company defining sovereignty as the thing institutions lose by depending on outside vendors is, in the same breath, urging governments to keep depending on its own American-built architecture rather than build their own, and dismissing the European push toward domestic alternatives as exactly the kind of politically driven decision-making it warns against.
That framing lands at an odd moment for Palantir’s actual footprint in Europe. Even as the company argues sovereignty means owning your own stack, France’s domestic intelligence agency, according to a report by Malay Mail on the AFP wire, renewed its Palantir contract for three years in December, with the agency itself noting the extension was granted only pending deployment of a homegrown replacement. The company’s sovereignty pitch, in other words, is being made to governments actively trying to exit the very dependency it describes, not adopt it.
That sovereignty argument is not just rhetorical; it maps onto a concrete legal strategy for winning contracts without competition. Palantir has repeatedly invoked, in a pattern tracked by Washington Technology, a 1994 federal law requiring agencies to buy commercial software over custom-built alternatives, using it to challenge a Defense Intelligence Agency solicitation this month just as it did in an earlier protest over the agency’s Prometheus project, and as it did in 2016 to force the Army to abandon custom development on its battlefield intelligence system in favor of Palantir’s own platform. Each invocation of “commercial-first” procurement law has expanded the company’s footprint inside a single agency at the expense of competitors, turning a rule meant to modernize government purchasing into a mechanism for consolidating one vendor’s position: the same institutional slowness the manifestos denounce is being used, case by case, to write the company deeper into the system.
Schmidt’s Self-Fulfilling Warning
Schmidt supplies the piece’s most systemic argument, and its clearest case of manufacturing the crisis it then offers to solve. Schmidt warned, in remarks to the Nuclear Threat Initiative reported by NTI, that the compression of decision-making time puts meaningful human control at risk, since today’s AI systems, however powerful, still make mistakes. What goes unstated in that framing is that the compression is not a natural phenomenon: it is a direct product of the autonomous drone and targeting systems that Schmidt’s own industry, including firms he has invested in, is racing to build and sell to militaries on both sides of every frontier. The warning and the product are the same technology, described twice.
At the Munich Security Conference, covered by ChannelDraw, Schmidt pressed the point further, arguing that 21st-century security will be determined more by artificial intelligence than by traditional weapons, and that whoever controls AI development will hold military superiority regardless of legacy arsenals. It is a claim that, not coincidentally, argues for concentrating exactly the kind of capability Schmidt’s ventures are positioned to supply.
Taken together, these positions describe less a shared ideology than a shared commercial mechanism. Musk treats geopolitics as a downstream effect of an accelerating technology curve that democratic debate cannot keep pace with, an argument that has helped normalize his own unaccountable influence over information and infrastructure. Karp treats restraint as a moral failure and casts his firm as the last honest broker of deterrence, while quietly using procurement law to lock in sole-source Pentagon business. Schmidt treats the compression of decision time as an argument for concentrating AI capability among a handful of powers and firms, having helped build the very systems doing the compressing. None of the three leave much room for the slower, more accountable version of security policy that European governments, NATO, and multilateral institutions still nominally practice, and each frames that slowness as the central danger of the moment, an assessment that happens to double as the sales pitch. Whether these figures are diagnosing a genuine erosion of state capacity or building the case for their own centrality is a question the record increasingly answers on its own: all three have direct commercial and political stakes in the outcomes they describe, and their warnings about institutional slowness arrive, with striking regularity, alongside proposals only they are positioned to fulfill.
