$7.7 billion is the projected market value of AI-driven defense technologies by 2028, a figure that underscores a critical tension: the creators of potentially world-altering AI are rapidly losing control over its deployment. The recent standoff between Dario Amodei, CEO of Anthropic, and the Pentagon isn’t an isolated incident, but a harbinger of a power dynamic where the imperatives of national security – and the logic of competitive advantage – will likely override the ethical concerns of even the most cautious AI developers. Follow the money, and the trajectory becomes clear: governments, facing geopolitical pressures, are prioritizing access and control over adherence to safety protocols, effectively nationalizing the future of artificial intelligence.
Amodei’s 15,000-word manifesto, “Machines of Loving Grace,” published over a year ago, envisioned a future where “polymath AIs” – surpassing Nobel laureates in key fields – would solve humanity’s most pressing problems by 2035. This optimistic projection, echoing similar sentiments from Demis Hassabis at Google DeepMind and Sam Altman at OpenAI, hinges on the assumption that AI development can be guided towards benevolent outcomes. However, the Pentagon’s recent ultimatum – demanding unrestricted access to Anthropic’s AI model, Claude, beyond existing legal constraints – reveals a starkly different reality. This isn’t about nurturing a “country of geniuses”; it’s about securing a military edge, even if it means sidelining the very individuals who understand the technology’s potential risks. The $7.7 billion figure isn’t just about innovation; it’s about weaponization.
The historical parallel to the Manhattan Project, frequently invoked by Amodei himself, is particularly apt. Just as the physicists who unlocked nuclear energy quickly lost control over its application, today’s AI researchers are discovering that creation doesn’t equate to control. The initial promise of “energy too cheap to meter,” as articulated by Lewis Strauss, chair of the Atomic Energy Commission, never fully materialized, and the utopian visions of reshaping landscapes with nuclear explosions – proposed by Edward Teller – remained largely unrealized. Similarly, the current hype surrounding “intelligence too cheap to meter,” as recently riffed on by Altman, risks obscuring the fact that the real-world application of AI is being shaped by forces far removed from the labs where it’s conceived. The contrast is telling: a projected $7.7 billion defense market versus the uncertain economic benefits of a truly democratized, universally beneficial AI.
This article draws on reporting from theatlantic.com.
The Pentagon’s coercive tactics – employing a supply-chain-risk designation that threatened Anthropic’s business – demonstrate a willingness to prioritize national security over the principles of responsible AI development. This move, unprecedented against an American company, underscores the government’s growing impatience with any limitations on its access to cutting-edge technology. Anthropic’s red lines – preventing mass surveillance of citizens and prohibiting autonomous weapons systems – were dismissed, highlighting a fundamental conflict between the values of AI developers and the strategic imperatives of the state. OpenAI’s subsequent swift agreement to a Pentagon deal, described by Amodei as “safety theater,” further illustrates this trend. The speed of the deal, and the perceived compromises made, reveal a willingness to prioritize access over caution.
The historical precedent is clear: the initial optimism surrounding nuclear technology gave way to a decades-long arms race and the constant threat of global annihilation. Despite proposals for international control – championed by figures like Niels Bohr and J. Robert Oppenheimer – the logic of competitive advantage ultimately prevailed. The expiration of the last remaining treaty constraining US and Russian nuclear arsenals last month serves as a chilling reminder of this pattern. This same logic is now poised to shape the future of AI, with nations vying for dominance in a technology that could redefine the global balance of power. The fact that the U.S. is already reportedly using Anthropic’s AI in sensitive operations in Venezuela and Iran, even before the technology is fully mature, demonstrates the urgency with which these tools are being deployed.
What this means for your wallet: expect continued investment in AI, but increasingly directed towards military applications. While consumer-facing AI products will continue to evolve, the true economic impact of this technology will be felt in the defense sector, and the ethical considerations will be increasingly overshadowed by geopolitical competition. The question isn’t if AI will be weaponized, but how – and whether the creators of this powerful technology will have any say in the matter. Watch closely for the next round of Pentagon contracts, and consider whether the pursuit of a military AI advantage is worth the potential risks to global stability and individual liberties.
