When Eric Wu left Opendoor in 2022, he had spent eight years steering one of real estate technology’s most ambitious companies through a market that rates then abruptly killed. A year later, convinced that artificial intelligence represented the defining platform shift of his lifetime, he went back to building. The result is NavigateAI, a startup whose bet on construction labour shortage has attracted $25 million in seed funding at a $225 million post-money valuation, led by Elad Gil.
‘I knew if I looked back in 10 years and didn’t do something related to it, I’d probably regret that,’ Wu said of his decision to re-enter company-building rather than slide into investing.
The Worker Gap Behind the Bet
The labour problem Wu is building into is real and worsening. Associated Builders and Contractors put the shortfall at roughly 349,000 additional workers needed in 2026 just to keep pace with demand, rising to 456,000 net new workers in 2027 as construction spending growth is forecast to resume. An ageing workforce, tighter US immigration enforcement, and a wave of large infrastructure projects are all squeezing supply simultaneously.
Data centres are the loudest symptom. Kelly, the global staffing company, has said that 90% of data centre operators now name staffing shortages as a critical constraint on their ability to build or expand. Where a large campus once peaked at around 750 workers, the biggest projects now operate at a different order of magnitude entirely. OpenAI’s Stargate site in Abilene, Texas has reportedly involved 6,400 workers. Meta’s Richland Parish campus in Louisiana, following a July 2026 expansion announcement that brought total committed investment to more than $50 billion, is now expected to support over 7,500 construction jobs at peak, according to Meta’s own data centre announcement. That figure supersedes the roughly 5,000 workers cited at an earlier project phase.
NavigateAI’s Construction Labour Play
NavigateAI, founded in 2025, is built around a simple interface: a construction worker points a smartphone camera, or the camera on Meta’s AI glasses, at whatever they are installing and asks a question in plain language. Is the torque correct? Does this meet code? The system pulls building specs, manufacturer manuals, and company policy in real time. Wu describes it as a hands-free expert coach for people physically building things.
The founding team draws on Opendoor, Stripe, DeepMind, Stanford, and Google, according to the Investing News Network. Launch partners include Lennar, Tishman Speyer, and Roofstock. The investor list is similarly well-connected: Gil led the round, with Khosla Ventures, Fifth Wall, Helix Electric, and Tishman Speyer participating, alongside angels including Tony Xu of DoorDash, Apoorva Mehta of Instacart, Coinbase CEO Brian Armstrong, Logan Green of Lyft, and Zach Frankel of Ramp, Dallas Tanner of Invitation Homes, and Winston Weinberg of Harvey.ai, among others.
Gil’s read on the opportunity was direct: ‘Eric is one of the best founders I’ve backed multiple times now going after one of the biggest unsolved problems in the economy. AI copilots for the physical world is the kind of application that will define the next era of AI.’
On pricing, NavigateAI has shifted away from a usage-based token model toward a share of value created. If the platform helps a builder reduce the all-in cost of a home from $300,000 to $280,000, NavigateAI captures roughly 20% of that $20,000 saving. Wu points to Lennar, which spends roughly $9 billion a year on labour, installation, and construction, as an illustration of the addressable value: a 5% to 10% efficiency gain would represent hundreds of millions of dollars.
The longer game is data. Every completed job generates labelled, egocentric video of workers building physical things correctly and incorrectly. Wu believes that dataset will eventually be as valuable to robotics companies as the software product itself is to construction clients.
The challenges are not small. Value-based pricing creates an attribution problem Wu openly acknowledges: proving a home was built faster because of NavigateAI rather than favourable weather or a strong crew requires A/B testing across divisions and is, by his own admission, approximate. Veteran journeymen remain resistant; Wu describes adoption curves splitting sharply between younger workers who embrace the product and experienced tradespeople who trust their own judgement. Safety liability is another open question: if NavigateAI clears a connection that later fails, defect liability in a litigious sector becomes NavigateAI’s problem too. Competitors including Buildots and OpenSpace already operate in adjacent territory, though Wu argues they are ‘more focused on project management’ while Navigate is built around ‘the individual labor’.
NavigateAI is also partnering with AIM, a Meta-backed fibre installation trade school that guarantees job placement to graduates, giving it a channel to reach workers before they arrive on site. Getting younger workers comfortable with AI-assisted work during training, Wu says, is where adoption habits form most readily.
Wu has no board yet, and says he is ‘going to try to go as long as I can without one,’ a posture that draws a clear line back to his years as a public company CEO at Opendoor after its 2020 SPAC listing. The question now is whether the proprietary data moat he is building accumulates fast enough, as LLM providers and well-funded rivals eye the same workforce gap.
