This is an excerpt from Shyam Sankar and Madeline Hart’s forthcoming book Mobilize: How the Reboot the American Industrial Base and Stop World War III. Mobilize is available to order now.
Everything about how Marine Colonel Drew Cukor ran Project Maven, the Department of Defense’s upstart AI initiative, put a target on his back. He infuriated the acquisition community, which is a powerful enemy in the Pentagon. Ultimately, the firestorm of criticism triggered a series of unfounded but unrelenting IG reports that would harry Cukor until his retirement. Some of the details that follow may seem obscure, but they’re essential to understanding the bureaucratic inertia and pettiness that hold our military back.
When Cukor launched Maven in 2017, the government still bought software like it bought hardware. This posed a problem. The phases of a hardware program are research, development, test, and evaluation (RDT&E), followed by production and sustainment. Costs are very high initially, and then they decline. The Department of Defense treated software the same way. It paid a lot up front for a systems integrator to build software, then it paid very little when the software went into production for patches and minor security upgrades. Software was treated as a static, finished product once it entered production.
Here’s the problem: software (at least, good software) is not static. It’s constantly improving, yet the cost is relatively flat across stages of development, which is why you pay a recurring subscription for commercial software instead of a large, upfront fee. This insight is the basis of the software-as-a-service model, and it enables constant improvement of the product. Development, testing, and production of software happen simultaneously, all the time. Understanding this, Cukor made the heretical argument to Congress that Maven should be procured as a continuously evolving capability, with a similar cost over its lifetime. Cukor procured software using Broad Agency Announcements (BAAs), a flexible contracting vehicle that categorized software as RDT&E. Although this categorization wasn’t perfect, the BAA allowed the program costs to reflect how software was developed and deployed and allowed Cukor to make frequent changes to the product while it was in production.
AMERICA MUST POWER AI WITH SPEED AND DISCIPLINE — OR CHINA WILL DOMINATE
Cukor would soon run into other problems with categorizing software as RDT&E. The department’s general posture is that if the US government is paying for R&D, it should own the intellectual property (IP) that results from that work. The problem is that despite the categorization of the contract vehicle as R&D, Maven wasn’t paying for commercial companies to perform R&D. When Palantir or Microsoft or Amazon showed up on day one of their work with Maven, they showed up with products that had decades and billions of dollars already invested. The R&D was already done. Yes, that product would get fine-tuned during the program and the companies would learn from the government’s mission and data, but fundamentally, the government was paying for software, not R&D. To Cukor, the government’s obsession with owning IP was an “overstated matter” more likely to harm the companies, and therefore national security, in the long term. As Cukor correctly notes, “If you [the company] can’t monetize this after working with us, then what’s the use of doing this? Why would you hand over your IP ever?”
To be clear, the companies did not own the government’s data and were not free to, say, sell a terrorist-targeting algorithm to China. International Traffic in Arms Regulations (ITAR) were in place, and the government’s interests were protected. But a company that built a deep learning algorithm maintained the IP to its proprietary model weights. For Palantir, this meant that we retained the IP to our core platform while giving the government rights to Maven-specific logic configured on top of it.
Safe to say, Cukor’s approach was correct. Almost a decade later, Maven remains the best example of a robust ecosystem of leading commercial technology companies working with the government. Unfortunately, Cukor’s view on IP remains in the minority. It was heretical then, and it’s heretical now. For this heresy, Cukor was cast by his enemies as acting against the interests of the government. “I was considered to be just a horrific human being.… There’s a whole class of people in the government that will go to their grave hating me because I would not compromise on this topic: platform IP belongs to the vendor, configurations on top are the customer’s.”
What happened next is almost hard to believe, if you know little about how the government operates: Cukor was punished for being too effective at his job. He was very good at rapidly getting money for Project Maven because he knew how acquisition worked and because his program was delivering. What’s more, he viewed acquisition as a form of “maneuver warfare” and never underestimated its importance as a source of continuous, rapid change to solve the most difficult problems.
MADURO RAID QUESTIONS TRIGGER PENTAGON REVIEW OF TOP AI FIRM AS POTENTIAL ‘SUPPLY CHAIN RISK’
In the Pentagon, the easiest way to attack someone is to accuse him of stealing money and issuing contracts illegally. For almost the entirety of Cukor’s time running Maven, a vicious stream of anonymous complaints were filed against Cukor. Some of these complaints were fueled by personal vendettas. It was a clear abuse of the process, but each allegation was treated with the utmost seriousness. Cukor was forced to face off against his mostly faceless opponents with little more than a heavily dog-eared copy of the Federal Acquisitions Regulation (FAR), the bible for procurement law and regulations. It had a permanent spot on his desk.
One day, the under secretary—Cukor’s boss—received an anonymous, five-page letter with a litany of terrible accusations against Cukor: he was corrupt, with bags of government money in his house that he used to buy expensive cars. He was wining and dining people to get contracts to move faster. His use of BAAs was illegal. He was setting himself up for a plush job after Project Maven. He had created a command environment that did not respect rank. (To this charge, Cukor pleads guilty: “I had some very strong captains that would happily tell off a colonel or general if they were wrong. We had a climate of moving fast and getting things done.”) Worst of all, the letter alleged, Cukor was illegally harboring a family of foreigners in his basement. This last, fantastic allegation came about because Cukor sponsored the (very legal) immigration of exceptional foreign mathematicians.
Cukor explains why he was a target: “You just have to understand this: when one group of people in the Pentagon get ahead of everybody else, the natural reaction is to kill that thing and get everyone back in line. That’s the Pentagon.” One is reminded of the Soviet Union, where the central government suppressed exceptional individuals who threatened the state’s uniformity and control. Everyone was doing exceptional work, which meant no one was.
Cukor told his boss the allegations were patently false and demanded the identity of his accuser. But his boss insisted on a full investigation. An Army officer was hired to investigate Cukor. This was a bad omen. The Marines and the Army have a long-standing rivalry that became even more acrimonious when the Army advocated abolishing the Marine Corps during the reorganization debates in and immediately after World War II. Harry Truman, partial to the Army, famously said that the Marines “have a propaganda machine almost the equal of Stalin’s.”
The Army officer published his investigation, but the best he could find, in his opinion, was that Cukor had not properly enforced rank, thereby creating a command climate that the Army officer said was anti-military. There were no allegations of criminal conduct. What he “found,” essentially, was that Cukor let his captains loose and didn’t enforce niceties—hardly fireable offenses. And what about the crazy allegations of money laundering and human smuggling? The Army officer didn’t have the skills to look into these matters, so he recommended that the Naval Criminal Investigative Service (NCIS) do it instead.
At this point, Cukor’s ordeal turned from tragedy to farce. When an NCIS investigator showed up at Cukor’s 1,400-square-foot home in Northern Viriginia, where he lived with his wife and four kids, there were no bundles of cash, fancy cars, or illegal immigrants in sight (although there were a few modest vehicles, all with more than 100,000 miles). The investigator left in disbelief. How had Cukor managed to support all these people on a government salary in such a small house?
PENTAGON’S AI BATTLE WILL HELP DECIDE WHO CONTROLS OUR MOST POWERFUL MILITARY TECH
That the NCIS found no incriminating evidence further enraged the establishment. Their options dwindling, they seized on a final chance to attack: Cukor’s retirement. After thirty years of exceptional service, Cukor had announced his intention to exit. Because of the baseless allegations, he knew there was no path for advancement. But instead of letting Cukor retire in peace, his critics went for his rank, threatening to demote him to lieutenant colonel!
At this point, any confusion on your part is excusable. Shouldn’t the Marine Corps be fighting for the person responsible for bringing AI to the Department of Defense? One of its own? Cukor finds the suggestion quaint. No, “the institution is always more important than the individual. We all know this; we sign up knowing this.” And Cukor was now associated, however baselessly, with money laundering, luxury cars, and undermining national security. He underwent two years of soul-crushing IG investigations that never really ended.
Cukor’s critics eventually gave up their campaign to take his rank, but he still suffered one final indignity on his way out the door. The last conversation that Cukor had before exiting the Pentagon was with the IG, who made clear that while Cukor was walking free today, the investigations would stay open for years. They could come after him at any point during that window.
STEVE FORBES: THE AI COLD WAR HAS BEGUN AND AMERICA CANNOT AFFORD TO LOSE
In 2022, after Cukor had retired, the Office of Inspector General finally published an unclassified but redacted version of its findings, “Evaluation of Contract Monitoring and Management for Project Maven.” The sanitized report contains no findings of fraud or impropriety. The primary conclusion is that Project Maven was indeed run “in accordance with FAR, DFARS [Defense Federal Acquisition Regulation Supplement], Defense Grant and Regulatory System, and contract requirements.” The worst the IG could find is that the “AWCFT did not document its approach to monitoring by formalizing the reporting metrics, processes, and procedures for monitoring and managing Project Maven contracts.” Cukor disputes even this one minor, critical finding. If you bother to read deeper in the report, it supports Cukor’s claim, too. Maven “actively monitored contract deliverables using AWCFT-developed reporting, metrics, processes, and procedures to meet Project Maven objectives,” and it scheduled “frequent and transparent programmatic reviews.” The IG admitted that monitoring and management techniques for AI and machine learning “are not captured in current procedures and best practices that are used by the DoD acquisitions community.” If only the IG applied such scrutiny and thoroughness to outcomes, rather than process. We should all be a little more concerned with whether a program actually works and a little less concerned with whether bureaucrats are checking the right boxes along the way.
The IG did, begrudgingly and in its own way, admit that Project Maven worked. It explained that documentation was needed, or else “future DoD acquisitions related to this complex, rapidly-moving technology may not benefit from the AWCFT’s monitoring and management lessons learned.” In other words, the IG criticized Maven for making it harder for other programs to learn from its example! The IG doesn’t write reports like this. It’s the equivalent of going before the Spanish Inquisition and coming away with a gold star for good behavior.
By the time the report was published, Cukor had already been driven out of the military. He’d had several chances for promotion, but because of the litany of accusations against him he couldn’t even be on the list of potential candidates. By the time his name was cleared, it was too late. What type of people do get promoted? Per Cukor:
Those that ascend are a rare breed: they’ve figured out how to survive in an environment where people can log any complaint against them and start investigations that jam up everything. This often results in a risk-averse senior leadership who avoid controversy at all costs. And the IG process is an unfortunate reality that favors the status quo and instills institutional complacency.
By contrast, Cukor had relentlessly pushed a contrarian AI agenda. People didn’t like it when a colonel ran through their organization at breakneck speed, delivering new technology via real-word experimentation, unorthodox contract terms, and vendors far outside the Beltway.
As Cukor recounts this vendetta, he does so without bitterness. There’s passion in his voice, but no anger. There’s no victim mentality. It’s actually kind of weird. Most people would, understandably, be bitter. Cukor attributes his equanimity to his Marine stoicism. He knows what’s right and what’s wrong. “There are many of us like that in the military. That’s why you have people who literally jump on hand grenades. They’ll do anything because it’s what’s right.” What’s more, the bad actions of others were often a source of motivation. This is the reason he was able to continuously deliver Maven even while these investigations were ongoing. After the fact, people on Maven were shocked to learn he’d been under investigation for more than two years, because it hadn’t altered his focus or output one bit. One engineer said that Cukor so effectively shielded the team from the politics that he had a nickname for him: the “iron dome of Pentagon bullshit.”
CLICK HERE FOR MORE FOX NEWS OPINION
Project Maven was the culmination of Cukor’s military career. Fighting for better intel methods and technology, fighting for Legacy to get police intelligence on the insurgencies in Iraq and Afghanistan, fighting against non-performant programs such as DCGS—all of these experiences trained him to bring a revolutionary AI effort to the military when the cards were stacked against him.
Significantly, Cukor was in his seat for five years—long enough for it to count. Too many talented officers are rotated in and out of their positions every two years. How many potential Mavens has the military lost due to constantly rotating personnel policy? Cukor is also a prime example of why you can’t separate the role of creating requirements from the role of delivering capabilities: designer and builder must work together. Much like Rickover built and then operated nuclear submarines, Cukor created the specifications for the AI solutions he wanted to exist, coordinated them, and then built them.
Cukor insists that while he and his team accomplished something exceptional with Maven, it need not be the exception. There are many others like him out there, just waiting for a chance and a climate that doesn’t presume they’re guilty until proven innocent. In many ways, Cukor views himself as a typical Marine: he came from a humble background, imbibed the service’s values, and put his training to good use.
Perhaps most important, Cukor is living, breathing proof that herculean effort and selfless service are still possible in government—even in as flawed and sclerotic an institution as the Pentagon. We think of titans like Rickover as existing solely in a bygone and inaccessible age. Cukor shows that isn’t true, either. Cukor had a book about the Yazidis, a basement office, and a righteous fire burning within him. That was enough for him to revolutionize the Pentagon and the way we fight wars forever.
Madeline Hart is a Defense Lead at Palantir Technologies, where she works on next-generation defense and space products. She started Palantir’s First Breakfast publication.
