Defense Secretary Pete Hegseth used a speech at Marine Corps Base Quantico on Wednesday, September 30, 2026, to announce Project Meridian, a 120-day Pentagon study of how the United States will fight future wars. Three outsiders lead it: Elon Musk, Palmer Luckey and former House Speaker Newt Gingrich. Two days later, the panel of the Moonshots podcast debated what it means. One member called it a transformative step. Others asked whether the Pentagon can move fast enough to use what it finds, and how much weapons should be allowed to decide on their own.
What was announced
Peter Diamandis, the founder of XPRIZE and the show's host, closed the episode, recorded October 2, with the story. He read Hegseth's description of the project: it is "not about developing new strategies or new policies," but about "discovering, developing, and fielding the weapons and systems future troops will need on the battlefield from the Earth to beyond the moon."
According to GovConWire's report, Musk, Luckey and Gingrich are co-directors, and the Pentagon's chief technology officer, Emil Michael, facilitates the effort. The study is meant to produce an actionable report on developing, testing and fielding future capabilities. An unclassified report and a classified annex are due January 28, 2027.
Diamandis said he likes commissions with a time limit that don't take Musk away "for, you know, a year at a time." He also noted that SpaceX and Anduril both hold multibillion-dollar defense contracts, and that Anduril is building autonomous weapons. GovConWire identifies Musk as founder and CEO of SpaceX and Luckey as a co-founder of Anduril.
"A transformative moment"
Alexander Wissner-Gross, a computer scientist and founder of Reified, was most enthusiastic. He was less interested in Meridian itself than in another initiative announced the same day: Project Agincourt, which he described as the standup of the Department of War's first autonomous-warfare command. "This, I think, is a transformative moment," he said. For the first time, he argued, the military would have a single joint force dedicated to autonomous weapons, including drones "but not exclusively drones," under a four-star functional combatant command. Combatant commands are the military's top-level warfighting organizations. Most cover a region of the world; functional ones are built around a single mission instead.
Wissner-Gross said that if he were secretary for a day, he would create several more such commands, and that this one would have been on his top-five list.
Agincourt is still a plan. GovConWire reports that the goal is to set up the command by October 1, 2027, with its own personnel, budget, acquisition and career structures. The Pentagon's unmanned-systems portfolio office is to lead the work, coordinate any legislation it needs with Congress and test a new way of buying equipment. An initial action plan is due within 30 days.
Wissner-Gross also pointed to a phrase in the Meridian announcement: it would examine "the full spectrum of future warfighting domains from subterranean depths to the cislunar frontier." Cislunar space is the region between Earth and the moon. He read the two ends of that range as the ocean bottom and lunar space, which he called "wildly underserved domains, not just for warfare, but for peacetime activities as well." (The GovConWire report describes the lower end as underground operations.) "We know embarrassingly little about our ocean bottoms," he said. If the department now invests in American exploration of the seafloor, the lunar surface and the space around the moon, he expects that to pay dividends for the whole economy and for technology more broadly, in ways that "might naively have nothing to do with warfare." Diamandis added that we know more about the surface of Mars than about our own ocean floor.
Who decides to kill
The guest, Richard Socher, co-founder and CEO of the startup Recursive and founder and CEO of You.com, was more cautious. Diamandis reminded him that his essay lists autonomous weapons among four genuine concerns about AI, and that Socher has said AI should never control lethal decisions without human oversight.
Socher said he stood by those points. Military AI is "not a particular area that I am excited about applying AI to," he said, though he expects people to pursue it anyway. "I'm not a doomer at all," he said, but giving AI access to all the nuclear weapons and connecting it would be "a really poor decision." "That's literally how Skynet and Terminator 3 get started." He would rather point superintelligence at scientific discovery. His rule of thumb is that the closer a decision comes to ending human lives rather than just affecting them, the more human oversight it should have.
Dave Blundin, founder of Link Ventures, was torn. He said he is "overjoyed" that capable people like Luckey and Musk are now willing to work in Washington, after a career in which trips there meant meetings with "lawyers and politicians and occasionally an accountant." But he said, "I'm also not a fan of autonomous weapons that make decisions in the field." Luckey had made a very good case for them in the panel's earlier podcast from Los Angeles, Blundin said: "I don't agree that it's a good choice, but he actually has some very rational arguments for why it's going to be that way."
Blundin passed on one of those arguments. Luckey had said that under the ocean it is not practical to communicate with a central server or command, so that work is already automated. "Once it goes into hunt mode, it just hunts and it's not communicating back," Blundin said. He added that he did not know what triggers that mode.
Can procurement keep up?
Salim Ismail, founder of Open ExO, said the hard part is not spotting future technologies, which "we can all see." It is absorbing them. The Pentagon is trying to rethink what he called a 100-year-old organization, and the open question is whether "a procurement organization built for 20-year weapons programs" can operate on "a 90-day technology cycle." Procurement is the system the military uses to specify, buy and field equipment. Exponential technology, Ismail said, cannot be bought with linear procurement. Agincourt's plan to test a new acquisition approach bears on the same problem.
To show how fast things are moving, Ismail gave drone figures from the war between Russia and Ukraine. Two years in, he said, the two sides were using about half a million drones. This year, by his figures, Russia and Ukraine will each make 10 million. He called it "an unbelievable escalation." He saw one upside: the two sides are increasingly fighting each other with drones, "without humans in the loop," though he acknowledged the drones are doing far more damage. His second statistic was about 10,000 drones a month crossing the US–Mexico border. Wall technology, he said, "is not as good as drone technology," so a border wall "is not the greatest idea right now."