In February, 24-year-old Ian Lafferty decided to take a break from the project he was working on—an AI tool for email—and attend a hackathon in San Francisco. While there, he met a handful of Ukranians fresh from the front lines, looking for new tech they could bring to the cause.
Ukraine has gotten around shortages in artillery by relying on cheap drones. But those drones require GPS guidance, which can be jammed.
So over the course of a couple days, with very little sleep, Ian and his two partners figured out a new system to allow drones to locate where they were, without having to rely on a signal beamed down from a constellation of satellites. The trick: have the drone’s cameras take pictures of the and compare those to a database of Google image maps, using simple machine learning. Within 24 hours, the team had cobbled together a GPS-independent drone for less than $500 dollars.
“You can fit a lot of compressed maps on, like, a 256 gigabyte SD card if you compress them the right way… We can fit, you know, 10,000 plus-square kilometers,” Lafferty told Defense One. Uploading the maps onto the small computer that can be affixed to virtually any drone frame takes just a couple of hours.
Lafferty and his partners at tTheseus are now working with a U.S. Army Special Operations group, Defense One confirmed, to test the drone in exercises and experiments. There’s still plenty of work to do to make sure the system works across a wider array of altitudes and speeds, but those are solvable problems, he said.
There have been multiple conversations with representatives from the Ukrainian military after that initial meeting, Lafferty said. And they’ve designed the system to accommodate for the fact that ground truth doesn’t always fit the static picture.
“I mean, stuff gets blown up all the time. Like, houses just disappear.. If you’re looking for a house and there’s no house, what are you going to do?” Lafferty said.
Theseus’s story reveals a lot about the Pentagon’s changing relationship with non-traditional companies and innovators. For example: breakthroughs in the defense space are no longer the sole domain of a handful of established defense contractors. Advances in AI—coupled with decades-long trends in information technology—are lowering the bar to bringing important new capabilities online. Now, a group of smart young people with no experience in the military can create new battlefield-relevant capabilities from cheap, easily available components, and do so at a fraction of the time and cost of a traditional defense contractor.
It also shows that the culture of young Silicon Valley programmers and aspiring company founders is changing. The taboo of collaborating with the Defense Department is not what it was in 2018, when Google programers objected so strongly to the company’s work with the Pentagon that the company dropped the contract.
That change is fueled in part by the success of a new breed of defense-focused startups like Palantir and Anduril, who are winning contracts and showing that digital-first processes in design and manufacturing can deliver new capabilities much more quickly than in the past.
“Thanks to the Pentagon, the lords of Silicon Valley are having a moment,” Gizmodo declared this week.
Y Combinator, perhaps the best-known startup accelerator in the world, used to be the place where young programmers went to launch consumer-facing companies like DoorDash, Instacart and Airbnb. This week, they announced the launch of a company called Ares Industries. The product? Low-cost cruise missiles.
But while startup culture is changing to become more Pentagon-friendly, the Pentagon still isn’t changing fast enough to give young defense companies the support they need to grow, says Michael Brown, a partner at Shield Capital and the former head of the Defense Innovation Unit.
“Unfortunately, procurement from venture-backed startups represents only 1% of the procurement dollars of DOD,” Brown told Defense One.
At Shield, Brown has a great view of the changing Silicon Valley landscape. The new openness to defense is real, he said.
“Engineers continue to want to work on important problems, and as the world has become a more dangerous place given the aggressiveness of Russia, China, and Iran, national security problems are among the most challenging, but whose solutions can have the highest impact.”
There are still big structural obstacles to better collaboration. For one thing, if a company wants to make tech that could be relevant both to consumers and the Defense Department, working with the Pentagon means rejecting a lot of foreign funding. which could go to a competing company with less patriotic ambitions.
Young founders aren’t ready for the vetting and scrutiny that Defense Department money brings, he said: “How could they be? Navigating DOD is still one of the most challenging of any customers, given that it’s literally hundreds of individual buying entities that must conform to thousands of pages of acquisition rules and guidance. We have a long way to go beyond successful efforts like the Defense Innovation Unit to make it easy to sell to DOD,” he said.
But there are small changes the Defense Department could make in the way it usually buys things, particularly from dual-use or consumer tech companies.
“When the Defense Department buys commercial items, which the law requires when they are available, there’s no need to undertake the traditional process of writing what the DOD would like to have built. Instead, there should simply be validating a need and beginning a selection process for vendors” Brown said.
The good news is, “there’s plenty of U.S.-based capital, given the current boom in defense tech, which has increased an order of magnitude in recent years and attracted specialist firms like Shield Capital as well as generalist firms like a16z and General Catalyst,” he said. “The obstacle for startups isn’t a lack of U.S. capital, it is the Defense Department’s lack of budget flexibility to rapidly shift appropriated funds to new technologies.”
Read the full article here