Ex-Opendoor Chief's AI Startup Bets Construction's Labor Crunch Is Its Opening

Nine months after Eric Wu, the co-founder and former chief executive of Opendoor, quietly began building a company to put artificial intelligence in the hands of electricians, plumbers and framers, the startup has disclosed new details about how it plans to make money and what it has learned from its first paying customers. The company, NavigateAI, emerged from stealth in May with $25 million in seed funding, and a report published Monday by TechCrunch laid out the fuller picture of a business that has shifted its pricing model, signed on trade schools and homebuilders, and begun collecting the kind of on-the-job video data Wu believes could eventually feed robotics companies.

NavigateAI's product is a smartphone- and smart-glasses-based assistant that watches over a worker's shoulder, in effect, as they wire a panel or run a duct. A worker can point a camera at an installation and ask whether it meets code, get a torque specification, or pull up a manufacturer's manual, with the system drawing on building specifications and company policies in real time. The company has built the tool to run hands-free through Meta's AI glasses, a partnership Wu has said he considers central to the product's long-term value, on the theory that field workers cannot stop to type on a phone while holding tools.

The Numbers

NavigateAI's seed round, led by the investor Elad Gil, valued the company at $225 million and drew participation from Khosla Ventures, Fifth Wall, the homebuilder Lennar, the developer Tishman Speyer and the electrical contractor Helix Electric, according to the company's own launch announcement. Angel investors included Tony Xu of DoorDash, Apoorva Mehta of Instacart and Brian Armstrong of Coinbase. The company has since moved away from a token-plus-margin, usage-based pricing structure toward value-based pricing that captures roughly 20 percent of the cost savings it can attribute to its software, according to the TechCrunch report.

The market NavigateAI is chasing is defined by a persistent gap between the workers the construction industry needs and the workers it has. The Associated Builders and Contractors estimated in January that the industry must attract 349,000 net new workers in 2026, a figure it expects to rise to 456,000 in 2027 as construction spending accelerates again. Separately, Wu has cited a wider estimate, telling reporters the country is short roughly 500,000 builders overall, a number he says is projected to keep growing.

"We have a shortage of people: we're short 500,000 builders and that number is projected to increase to two million," Wu has said of the labor gap driving demand for the product.

How We Got Here

The shortage NavigateAI is selling into has been sharpened by the buildout of AI infrastructure itself. A single data center project once required a few hundred construction workers; now, as hyperscale campuses built to house AI training clusters have grown, staffing needs for individual sites have climbed into the thousands. Meta's own account of its Richland Parish, Louisiana campus, known as Hyperion, describes a construction workforce expected to peak at roughly 5,000 workers on a single site, up from about 3,700 currently on the job. OpenAI's Stargate data center project has reported comparably large crews, according to the TechCrunch report.

That demand has collided with an already strained trades workforce. A salary and hiring survey released in August by the staffing firm Kelly Services found that 90 percent of data center operators cite staffing shortages as a critical constraint on their ability to build or expand facilities, and it projected that roughly half of the U.S. data centers scheduled for delivery this year face delays tied at least in part to workforce planning. Wu, who built Opendoor over eight years before stepping down as chief executive in 2022, has said he sat out the following year before concluding that AI's disruptive potential in physical-world industries was too significant to ignore.

Who Is Affected

NavigateAI's early customers span homebuilding and commercial development. Lennar, Tishman Speyer and the rental-home operator Roofstock were named as launch partners in May, alongside a partnership with AIM, a Meta-backed trade school that trains fiber and electrical installers and offers job-placement guarantees, according to an earlier report on the company's stealth launch. The company has said adoption has broken along generational lines: younger workers have embraced the coaching tool readily, while veteran tradespeople with decades on the job have been more resistant to being guided, and monitored, by software.

The startup also faces open questions about liability and measurement that will shape how contractors and insurers treat the tool. It remains unclear, for instance, who bears responsibility if the software signs off on an electrical connection that later fails, and the company has acknowledged it is difficult to prove that improvements in speed or quality on a job site stem from its coaching rather than weather, crew turnover or material delays. NavigateAI competes for attention with platforms such as Buildots and OpenSpace, though Wu has drawn a distinction between those tools' focus on project-level monitoring and his company's emphasis on individual workers.

What People Are Saying, and What Comes Next

Wu has been candid that the video his software collects from job sites, footage of correct and incorrect installations captured from a worker's own vantage point, could eventually be as valuable to robotics developers as the coaching software is to contractors today. He has also structured the company without a board of directors for now, saying he intends to "go as long as I can without one" so the company can stay focused on customers rather than governance overhead.

• NavigateAI's seed round: $25 million at a $225 million post-money valuation, led by Elad Gil
• Construction workers the industry needs in 2026, per ABC: 349,000
• Peak construction workforce at Meta's single Hyperion data center site: about 5,000
• Data center operators citing staffing as a critical constraint, per Kelly Services: 90 percent

The company has not disclosed customer counts, revenue or a timeline for expanding beyond its current partners. Its bet is that a persistent, worsening labor shortage, driven in large part by the same AI data center boom that companies like NavigateAI are themselves part of, will make contractors willing to pay for software that helps fewer workers do more. Whether that bet pays off will depend on questions the company has only begun to answer publicly: how insurers and courts eventually assign liability for AI-assisted work, whether experienced tradespeople come to trust the tool, and whether NavigateAI can maintain a foothold as larger AI companies with their own hardware ambitions look at the same physical-world data it is racing to collect.