Early last year, Raj Shah and Akhil Iyer were in Hawaii meeting with the man entrusted with protecting American defense interests across 14 time zones when they heard a request for a startup.
As the leader of the U.S. Pacific Command, Admiral Samuel Paparo is responsible for a fleet of about 200 ships and hundreds of thousands of personnel, from Indian Ocean waters where his fleet has faced off with Iranian ships, to waters off Taiwan. TIME recently called him “The Admiral Who Must Prevent War With China” in a September cover story.
And as Admiral Paparo met with the investors from Shield Capital, a defense-focused venture capital firm, he highlighted a particular need: logistics, and logistics planning. Despite recent advances in AI and data software, important troop or ship movements still got decided in Excel spreadsheets, then emailed around.
It was a problem that resonated with Shah, Shield’s founder and a former U.S. Air Force fighter pilot, and Iyer, a combat veteran of the U.S. Marine Corps’ special operations command who served tours in Iraq.
“There was a very clear signal and demand from the partner warfighter,” Iyer says. “We don’t have the resources that we would like to be able to defend across this space, so we have to optimize as best and as fast as possible.”
But when they looked around for an investment to bring back to the admiral, they couldn’t find one. So they decided to build it.
Nearly two years later, that startup is decloaking out of stealth: Watchtower Labs. Incubated out of Shield Capital with Iyer as its CEO, Watchtower launches today with $35 million in funding co-led by Shield and Conviction, with participation from U.S. government-linked In-Q-Tel (IQT), and Founders Fund.
What Watchtower is building: a software suite that combines advances in AI with in-house simulation work to help defense planners make critical logistical decisions.
Users – picture the officers who report up to senior command with recommended actions or contingency plans – ask questions in natural language, and Watchtower’s locally run software uses LLMs to find and process the corresponding data, to best understand the scenario and its context. Watchtower then runs simulations by the thousands to return likely contingencies or suggested actions via a custom dashboard or an easily sharable artifact, like a presentation.
What happens if anti-air missile supplies run short, or there’s a cyber disruption? What does protracted conflict with Iran mean for capabilities in a hypothetical escalation with China?
“We’re building a product that peers into the future, at the distribution of outcomes that could be,” Iyer says. “Ideally, it’s early enough where you can do something about it.”
With a distributed team of nearly 30 across the Bay Area, Hawaii, Seattle, and Washington, D.C., Watchtower already has several early, prototype-style contracts with major operational commands, the startup says, as well as several other pilot customers that Iyer says he has to keep even more vague.
🎧 The Upstarts Podcast 🎧 is back, and last week we chatted with Neko Health CEO Hjalmar Nilsonne about his novel 1-hour, full body health scan. Check it out on YouTube, Apple, Spotify or wherever you get your podcasts.
Against a backdrop of rising geopolitical tensions and record venture investment in defense startups, Watchtower is interesting for several reasons.
First, there’s its technological approach: leveraging the latest in AI, but not a pure AI play itself, that slots in alongside (but purports not to compete directly with) other tools popular with defense buyers, like Anduril or Palantir. “You cannot LLM your way to supply chain optimization,” says Iyer.
Second, there’s the urgency that Iyer describes his prospective buyer as feeling. Resources feel scarce, he says. And while he doesn’t name specifics, recent reports have detailed warnings of missile shortages and a historically long deployment for a U.S. aircraft carrier, the USS Abraham Lincoln. Ukraine, meanwhile, reportedly ran out of munitions for its U.S.-made Patriot air defenses earlier this month.
And that’s all with China’s Pacific ambitions still mostly looming. Which means that for Watchtower, 2027 should prove a serious test.
More on Watchtower’s mission – and how its tech fits in the current AI-driven tech landscape – below.
A MESSAGE FROM OUR SPONSOR
Ship pricing like a product
Different product bets have different business goals. You’re either chasing market adoption, revenue growth, or profits. You need to test fitting pricing models for each of those goals. Fast. Chargebee can help — Price for usage credits, outcomes, complex contracts, or any other modern commercial construct.
Operation Watchtower
Watchtower is named after a military operation famous among Marines like Iyer: Operation Watchtower, a codename for the Guadalcanal campaign during World War II.
The long battle was an early test for American forces before the U.S. had built up its industrial muscle, as Iyer describes it, and one in which fog of war – limited visibility and intelligence on the ground – meant that leaders had incomplete understanding of their supply lines, down to how many bullets their troops on the island had left.
It’s a safe, classified- data-free metaphor for a defense logistics startup, one that doesn’t get into more sensitive personal experiences of Iyer or his customers. A Stanford graduate, Iyer says he always wanted to work in national security; after Stanford, he spent seven years in the Marines, including serving in counter-insurgency operations on deployment in Iraq.
While at Harvard Business School, he met Shah, another member of what Iyer jokes is a tight-knit Indian-American military community; following an internship stint at Anduril, Iyer joined Shield Capital, Shah’s firm, in 2021 to invest in defense and dual-use startups.
After the idea for Watchtower germinated following the fateful meeting with Admiral Paparo, it was natural that Iyer would lead Shield’s first incubation, according to Shah.
“He’s a multi-dimensional leader that’s excelled in all walks of life, so we’re super excited for him to be leading the Watchtower effort,” says Shah, who notes that Iyer remains active as a venture partner with the firm.
The move also made sense to Conviction’s Sarah Guo, who was working with Iyer on the board of a mutual investment, Distributed Spectrum, which is building AI-powered communications sensors.
“Humans in Excel and email, that is not a reasonable system in the age of superintelligence,” Guo says. “Akhil was like, ‘I think I’ve got to go do this,’ and I was like, ‘Okay, where do I sign?’”
More than an AI tool
With the funding, Watchtower plans to hire more, both for engineering and for working closely with customers. Veteran service status isn’t a requirement, says Iyer, although employees are expected to believe in the importance of the mission.
Such allies include the startup’s technical leaders, CTO Suren Aghajanyan and head of engineering John Morman, who both worked long stints as managers at Microsoft.
“I worked on a lot of cool stuff there, but had never done anything that felt really impactful in the sense that it would have a difference on the men and women serving our country, providing real value in that sense,” Morman says.
From a technical standpoint, Watchtower’s breakthrough is in how its in-house discrete-event simulation (DES) software works with the LLMs it uses to understand the customer’s request, Morman explains. “That’s where a lot of the power comes from, being able to describe without having to get into the detail on everything, but articulate exactly what I would want to have happen,” he says. “And then have the system go execute that.”
Its ongoing challenge: the messy, often secret and restricted data that can hold their customers’ answers. “We are surprised every time,” Morman says. “It’s supposed to become less surprising, but every single time we ship a feature, there’s a new unique way that we find that the data can be bad.”
That challenge is greater when customers are using Watchtower in their own secure environments, meaning that improvements to the company’s own software can lead to multiple versions or capabilities being out in the world.
Watchtower’s answer for that: forward-deployed legwork. Morman chats with me from Hawaii, where he and much of Watchtower’s team are set up this week on an exercise with an unnamed pilot customer.
What about the big AI labs like Anthropic and OpenAI, whose government contracts have made headlines and led to a high-profile clash between the former and the Department of War about “supply chain risk” that is currently being litigated in court?
Both the startup and its backers insist that while AI advances helped unlock the possibility of a tool like Watchtower, they’re only a piece of the startup’s solution. Guo, brought on in part to help the startup navigate frontier developments in AI, argues that Watchtower’s product is unlikely to become a point of focus any time soon.
“Do I think that it’d be impossible for the labs? No. Do I think that they’re going to do it any time soon? Absolutely not,” she says. “There are companies where I think about the labs every day, and this is one where I think about them never.”
The company has no immediate plans to move beyond defense, but Shah notes that Watchtower’s tools could eventually prove valuable in other regulated industries. Over the next year or two, Watchtower will look to succeed and expand in its pilots, Iyer says.
“Next year is a really big year. Both geopolitically and for us being able to keep momentum as a business.”