Xingxing Wang spent three hours fielding questions from institutional investors on August 7, and the message was blunt: Unitree Robotics has working humanoid prototypes, but the company is not making money from them yet. The Chinese robotics manufacturer, known for its quadruped robots that undercut Boston Dynamics on price, used its IPO roadshow to acknowledge what many in the industry already suspected—humanoid robot commercialization remains stuck between proof-of-concept and profitable deployment. Wang's management team confirmed that revenue from humanoid platforms would not materialize in meaningful volume until the final quarter of 2026 at the earliest, with 2027 looking more realistic for sustained orders. That timeline matters because it reveals how wide the chasm remains between viral demonstration videos and products customers will pay to deploy at scale.
Unitree built its reputation selling quadruped robots at price points that forced competitors to rethink their margins. The Go2 launched at $1,600 for the base model, a fraction of what Boston Dynamics charges for Spot. That aggressive pricing strategy brought the Hangzhou-based company attention and market share in inspection, security, and research applications. Quadrupeds brought cash flow. Humanoids, Wang made clear, do not yet. The roadshow discussion centered on the G1 humanoid, a platform Unitree has demonstrated in warehouse and light manufacturing scenarios, but the company has not disclosed how many units are in paid deployments versus evaluation pilots. Wang described current humanoid projects as collaborative development agreements with enterprise partners, a structure that typically means the robot maker is subsidizing deployments to gather data and refine the platform. That approach works when investor capital is available. It does not work if public market shareholders expect quarterly revenue growth.
The IPO timing raises questions about capital strategy. Unitree is going public while its most hyped product line—humanoids—remains pre-revenue, a structure more common in biotech than hardware. Wang and his team emphasized during the session that quadruped sales continue to grow and provide a foundation, but institutional investors wanted specifics on when humanoid units would transition from pilot programs to purchase orders. The management team outlined technical hurdles still being addressed: battery life in eight-hour shifts, manipulation tasks that require sub-millimeter precision, and the software integration needed to operate alongside existing warehouse management systems. Those are solvable problems, but they require engineering time and customer patience. The roadshow suggested that several pilot partners are waiting to see sustained performance over months, not weeks, before committing to volume orders. One investor asked directly whether any customer had signed a contract for more than ten humanoid units. Wang declined to provide a number.
The Unitree disclosure aligns with a broader pattern across the humanoid sector. Figure AI raised $675 million in a Series B round earlier this year and has placed robots with BMW, but the automaker has not disclosed how many units are operational or whether they have replaced human workers in any production cell. Apptronik announced a partnership with Mercedes-Benz in 2025, yet the arrangement remains a pilot with no public purchase commitment. Tesla continues to showcase Optimus in controlled settings, but outside observers have not seen the robot performing work in Tesla's own factories at scale. Sanctuary AI, Agility Robotics, and others have demonstrated impressive capabilities in narrow tasks—bin picking, box moving, simple assembly—but none have published data showing cost savings that justify the capital expense and integration complexity. Unitree's candor during the roadshow is unusual. Most competitors let investor enthusiasm outpace operational reality. Wang's willingness to set expectations lower may reflect the scrutiny that comes with a public offering, or it may signal that the company learned from the quadruped business that credibility matters more than hype when customers are writing checks.
The humanoid robotics market is not failing. It is maturing slower than the demonstration videos suggest. Unitree's roadshow comments point to a sector still working through the basics: endurance, reliability, return on investment. Engineers at logistics companies and manufacturers want proof that a humanoid can work a full shift, five days a week, for six months without constant recalibration. They want error rates low enough that supervision costs do not erase labor savings. They want integration simple enough that existing IT staff can manage deployments without hiring robotics PhDs. Those requirements are reasonable, and they are not yet met at scale. Wang's timeline—late 2026 for initial revenue, 2027 for volume—assumes that Unitree and its competitors will solve these problems in the next twelve to eighteen months. That assumption may prove optimistic. The gap between a working prototype and a product customers trust to run unsupervised is often longer than the gap between concept and prototype.
What to Watch: Unitree's IPO pricing and first-day performance will indicate whether public investors share Wang's patience or expect faster humanoid monetization. Track whether Figure AI or Agility Robotics disclose unit economics or deployment scale from their automotive and logistics partnerships before year-end 2026. Monitor whether any customer—particularly in warehousing or light manufacturing—publicly commits to a humanoid order exceeding fifty units, a threshold that would signal confidence in reliability and ROI. Finally, watch for Unitree's Q4 2026 earnings call; any revision to the humanoid revenue timeline will clarify whether the IPO roadshow set realistic expectations or simply bought the company time.
