Agility Robotics Lists on Nasdaq as Humanoid Robots Shift Focus to Conversion Metrics

America's first humanoid robot listed company, yet it happened to coincide with Wall Street's most skeptical day regarding AI.

On November 5, Agility Robotics landed on the Nasdaq, with the stock code AGLT. That day, U.S. stocks encountered a sell-off led by falling AI stocks, and the Nasdaq index fell by 5.5%, creating the worst single-day performance since April 2025. Agility once fell below its opening price during the session, but finally closed at 11.51 USD, up 2.2%.

On the day when AI stocks fell collectively, the market still gave this robot company, which has only 174000 USD in half-year revenue, a bit of face. However, this enthusiasm did not last long; as of November 9, Agility's stock price had already fallen back to 8.75 USD.

Capital is willing to give it a chance, but has not yet given an answer.

Agility was founded in 2015, originating from the robotics laboratory at Oregon State University. Its core product, Digit, is a bipedal humanoid robot, currently primarily used for moving cartons in warehouses and factories. Rather than dancing and doing backflips, Agility prefers to emphasize one thing: Digit is already doing work.

As of before the listing, Digit has accumulated over 65,000 hours of operation, moving over 100,000 turnover bins in a GXO warehouse, with an accuracy rate of approximately 98% when executing tasks; the company also has 9 customer facilities with committed deployments, as well as over 300,000,000 USD in Digit 5-year orders.

On the other hand, Agility's 2025 revenue was only 1,780,000 USD, with a net loss of 138,000,000 USD; in the first half of 2026, revenue fell to 174,000 USD, and the net loss expanded to 362,000,000 USD. The latter loss figure includes a large amount of equity incentive expenses, but during the same period, operating activities still saw an outflow of 86,200,000 USD.

Compared to 625,000 USD in the same period of 2025, it decreased by 72%. The prospectus attributes the reason to a one-time robot sale of approximately 500,000 USD in the first half of 2025, which did not occur in the same period of 2026.

Ran for 65,000 hours, obtained a 300,000,000 USD order, but the revenue in the first half of the year was only 174,000 USD. What Agility needs most now is to truly turn the first two numbers into revenue.

However, Agility's listing also put a rare complete commercialization funnel, which can be checked quarterly by the public market, into the spotlight. In the past, the industry compared who left the laboratory first, who entered the factory first, and who obtained more orders; now, investors will continue to ask: how much of the running time turned into paid revenue, how much of the orders turned into actual deployments, and whether it can break even and be renewed after deployment.

Humanoid robots have begun to calculate “conversion rates.”

1.

65,000 hours, first must become paid hours

In 2024, GXO signed a multi-year Robot as a Service agreement with Agility, and Digit entered a warehouse in Georgia from then on, responsible for moving turnover bins onto conveyor belts. Compared to most humanoid robots that are still in the video and pilot stages, it indeed entered continuous operating customer sites earlier.

But having operated does not equal forming commercial scale.

In Agility's 2025 revenue of 1,780,000 USD, about 60% came from related parties; in the 174,000 USD revenue of the first half of 2026, related parties contributed about 70%. The company has already accumulated real-scene experience, but has not yet converted these experiences into large-scale third-party customer revenue.

From the perspective of the revenue structure, Agility has not yet entered the RaaS subscription volume growth period at this stage. In 2025, approximately 1,550,000 USD comes from robot sales, accounting for nearly 87% of total revenue; in the first half of 2026, there is no confirmed robot sales revenue.

The first conversion hurdle is stuck here: a robot completing a task does not equal a customer being willing to continue paying for it. Pilots first verify safety, stability, and process compatibility; only when the robot stays powered on daily, completes enough work, and is more cost-effective than manual labor or traditional automation will it enter the customer's regular budget.

Safety fences once directly blocked Digit's expanded deployment at Amazon.

In 2023, Agility sent two Digit 4s into an Amazon warehouse south of Seattle to test the recovery and transport of already emptied bins. The robots were able to complete the tasks, but could only operate in areas isolated from employees.

Agility co-founder Jonathan Hurst recalled that when both sides discussed expanding the deployment, Amazon brought up its own wheeled robots' safety certification experience spanning three to four years and questioned how a robot that walks by dynamic balancing could pass this hurdle.

This also explains why Digit 5 put “can work next to people” as the first priority for product upgrades. Reducing the reliance on traditional safety fences is not just adding another parameter, but the prerequisite for whether the robot can enter more workstations and undertake more tasks.

Agility CEO Peggy Johnson said that companies do not buy technology, they only buy solutions to problems. 65,000 hours prove that Agility has already obtained learning data; how much of that is paid operating time, how many robots it was generated by, and whether the revenue per unit can grow, determines whether this data eventually becomes a business.

2.

A 300,000,000 USD order, staked on a robot that has not yet been delivered

Tearing apart the prospectus reveals that the 300,000,000 USD all comes from a related party customer, corresponding to 1,000 units of Digit 5 and a three-year RaaS contract. Whether the robots can be deployed depends on whether Agility reaches the agreed product functions, technical specifications, and contract milestones; the accompanying warrants obtained by the customer also vest progressively with the actual deployment of the robots.

In accounting, this order has not yet become revenue.

The reason is that the Digit 5 purchased in the order has not yet officially launched. It is planned to be delivered to early customers in the first half of 2027, and will only be fully launched at the end of the year. Agility promises that it can repeatedly move 50-pound heavy objects, only requires 9 minutes of charging after working for 90 minutes, can change end tools, and reduces reliance on traditional safety fences, working next to people.

These upgrades all point to the same thing: turning a robot from only moving turnover bins into one that can flow between multiple workstations and multiple shifts. Only when tasks increase and daily working time lengthens can the RaaS math hold up.

This constitutes the second conversion: from contract amount to actual deployment and recognized revenue.

Agility originally expected to obtain over 620,000,000 USD through a SPAC transaction and accompanying financing; after shareholder redemptions, the final fundraising was approximately 465,000,000 USD. For a company that treats RaaS as its primary commercialization path, this 465,000,000 USD is first a bridge fund. The robots must first be built, delivered to customer sites, and continuously maintained before revenue is gradually recognized over the coming years.

Agility calculated a bill on Investor Day: a single Digit running on RaaS for five years is expected to generate approximately 500000 USD in revenue, with a product-level profit margin of about 50% over five years, and hardware investment recovered in about 1.5 years. According to the company's calculations, if annual production reaches 1000 units, the product-level profit margin can rise to about 70%, and the hardware investment recovery cycle shrinks to within one year; reaching 10000 units, the recovery cycle might only be pressed down to within half a year. This is still a single-unit model that does not include the company's R&D and management costs.

How many robots, how many facilities, and how much recurring revenue 300000000 USD can convert will be more important than the order figures themselves.

3.

Expanded to three warehouses, GXO still hasn't calculated the return

The customer most worth watching for Agility is not that unnamed 300000000 USD buyer, but GXO.

In August this year, this global large-scale contract logistics provider gave an answer that was not very pretty: GXO has already conducted 45 humanoid robot pilots with multiple suppliers in total; according to GXO's own words, the number of humanoid robots officially put into production in 2026 is still zero, no return on investment has been obtained yet, and it may still be a few years away from ROI.

Two months later, GXO decided to expand Digit to two warehouses in North Carolina and Tennessee. The new project plans to start with Digit 4 in early 2027, followed by a shift to Digit 5.

Without breaking even, why continue to expand points?

Because what GXO is purchasing at this time is the learning curve of humanoid robots. The first warehouse verifies whether the robots can do the work, and the second and third warehouses verify whether the same set of capabilities can be replicated. Only if the new project can shorten the deployment cycle, reduce on-site engineers, and reuse original programs and operation and maintenance experience will the business model begin to show scale effects.

This is the third conversion: whether deployment can turn into ROI, renewals, and expansion.

At this step, robot companies can no longer just announce how many factories they entered. The market needs to see the recurring revenue, utilization rate, and throughput of a single robot, and also need to know how much cash is consumed to deploy one robot, and whether the second factory is cheaper than the first.

The SPAC transaction gave Agility a pre-listing equity valuation of 2,500,000,000 USD. What the market priced was clearly not the 174,000 USD revenue, but the future conversion rate: how much of the 300,000,000 USD in orders can be fulfilled, whether the 9 customer facilities can expand to more factories, and whether the 65,000 hours of data can make the next deployment faster.

This set of questions is equally valid for Chinese robot companies.

In August this year, Unitree landed on the Sci-Tech Innovation Board. It sold 5215 humanoid robots in 2025, achieving 868000000 yuan in related revenue, which has already proven that robots can be manufactured at scale and sold; however, in the first three quarters of 2025, its humanoid robot revenue still had 73.6% coming from scientific research and education, with the industry application proportion being 9.01%.

Agility and Unitree represent two different starting points. Agility has fewer robots, penetrates deeper into customer sites, and needs to convert operational depth into revenue; Unitree has larger shipments and more mature hardware sales, and needs to convert installation volume into robots that truly work every day.

After Unitree went public, it once rose by more than five times, then fell significantly from its peak. According to Reuters citing The Information, regulators have begun to slow down the IPO process for some humanoid robot companies, questioning how much real demand lies behind the high valuations. The attitude of the capital market has already changed: being able to run, jump, and ship products is still important, but whether these numbers can eventually translate into customer repeat purchases and sustained revenue is beginning to determine how long the valuation can hold.

So, rather than treating Agility as a unified valuation ruler, it should be viewed more as a “conversion table.” For humanoid robot companies, the next three most important conversion rates are pilot-to-paid, order-to-delivery, and delivery-to-renewal, ultimately landing on cash flow.

When humanoid robots begin to calculate conversion rates, entering the factory is just an entry ticket. Turning the work done by the previous robot into the revenue of the next robot is the real competition after listing.