Agility Robotics Is Going Public at $2.5 Billion. Its Robots Cost 18× More Per Task Than a $1,900/Month Alternative. The Bet Is on Versatility.

Two point five billion dollars. That is what Wall Street says a company is worth when it has sold approximately one hundred robots, has never disclosed revenue publicly, and operates a factory running at roughly one percent of its designed capacity. On June 24, Agility Robotics announced a definitive merger agreement with Churchill Capital Corp XI, a blank-check company backed by veteran dealmaker Michael Klein, to go public under the ticker AGLT in what the press release calls a transaction that will create “the only U.S. publicly listed pure-play humanoid company with proven, active commercial deployments.”

Shares of CCXI jumped 18% in premarket trading. Reuters noted the deal will provide more than $620 million in gross proceeds, including $200 million from a PIPE led by Foxconn. Investors include NVIDIA, Amazon, SoftBank Vision Fund 2, and Playground Global. Agility claims a total addressable market of roughly $1 trillion across manufacturing, distribution, and logistics environments in the United States alone.

One trillion dollars. A hundred robots deployed. Those two numbers, sitting next to each other in the same investor presentation, capture the central tension in humanoid robotics better than any thesis statement ever could, and the math to understand that tension fits on a napkin.

The Napkin Math Nobody Published

The question investors should be asking is not whether humanoid robots can move totes in a warehouse. That question was answered in January 2026, when UK startup Humanoid deployed its HMND 01 Alpha at the Siemens Electronics Factory in Erlangen, Germany, for a two-week operational trial. The wheeled humanoid robot autonomously executed tote-handling tasks (picking, transporting, placing containers for human operators) and hit every target metric: 60 tote moves per hour, more than eight hours of continuous runtime, and autonomous pick-and-place success rates above 90%. Siemens called it a “real-world milestone.”

That trial provides something the humanoid sector desperately lacks: a published throughput number for an actual robot doing actual work in an actual factory. So we can finally run the calculation that matters.

What does it cost to move one tote using a humanoid robot, compared to a human worker, compared to a purpose-built mobile robot? We gathered publicly available pricing, wage data, and operational benchmarks and built the comparison nobody else has.

Cost Component	Humanoid (Digit-class)	Human Worker	Purpose-built (Brightpick)
Capital / wage cost	$250,000 unit cost	$18/hr base	$1,900/month RaaS
Annualized cost basis	$83,333 (3-yr depr.)	$36,000 (2,000 hrs)	$22,800
Maintenance / benefits	$37,500 (15% CAPEX)	$12,600 (35% burden)	Included in RaaS
Insurance, training, turnover	$5,167	$5,000	N/A
Total annual cost	$126,000	$53,600	$22,800
Throughput	60 moves/hr	45 moves/hr (est.)	75 picks/hr
Annual operating hours	4,000 (2 shifts)	2,000 (1 shift)	8,760 (24/7)
Annual totes/picks	240,000	90,000	657,000
Cost per task	$0.53	$0.60	$0.035

Three inputs and a division sign. The humanoid, running two shifts at $250,000 purchase price with a three-year depreciation schedule, comes out to roughly $0.53 per tote move. A human warehouse worker at the national average of $17.98 per hour plus a 35% benefits burden costs about $0.60 per tote. And a Brightpick Autopicker 2.0, a purpose-built mobile robot running 24/7 on a robotics-as-a-service subscription, handles each pick for three and a half cents.

Read that last number again. Three and a half cents per task. The humanoid costs 15 times more. Not 15% more. Fifteen times.

Why the Numbers Look Even Worse on Close Inspection

Our humanoid estimate is generous. Digit’s actual battery specs, documented by Sacra, show two to three hours of heavy-use runtime before the robot returns to autonomous charging docks, with lighter-task endurance stretching to eight hours. The 60-moves-per-hour benchmark comes from the Siemens trial, which tested the HMND 01 Alpha, a wheeled humanoid, not a bipedal one like Digit. Wheeled platforms are mechanically simpler and more energy-efficient than legs. Digit walks on spring-loaded bird-like legs that trade energy for the ability to handle ramps, curbs, and dock plates, and there is no published Digit throughput number to substitute.

So the $0.53 figure assumes two-shift operation with no unscheduled downtime, humanoid throughput matching a wheeled robot with fundamentally different locomotion, and a three-year asset life that many industrial robots exceed. Switch to single-shift operation and the number jumps to $1.05 per tote. Assume Digit’s actual heavy-use battery life of 2.5 hours and the cost climbs higher still.

The human worker estimate is equally generous, but in the opposite direction. It uses the national average wage of $17.98 per hour. In the logistics corridors where Agility’s customers actually operate, from metro Atlanta (GXO) to Ontario (Toyota Motor Manufacturing Canada) to the German Rhineland (Schaeffler), warehouse wages are climbing past $22. According to KPI Solutions, 76% of logistics facilities now report workforce deficits, and economic forecasts point to 3-4% annual wage growth continuing through 2027. At $22 per hour with overtime in a tight labor market, the human cost per tote approaches $0.80 to $0.90. The humanoid starts to look competitive with workers. It never looks competitive with the Brightpick.

So Why Would Anyone Buy One?

Because cost per tote is the wrong metric for valuing Agility. The company knows this. Every slide in the investor presentation that mentions “general-purpose” is telling you the same thing: Digit is not supposed to be the cheapest way to move a tote. It is supposed to be the only robot that can move a tote and then walk over to a different station and do something else entirely, in the same building, during the same shift, without any facility reconfiguration.

A Brightpick Autopicker moves totes. Magnificently, cheaply, endlessly. But if the warehouse reorganizes its aisles, or the customer switches from totes to pallets, or a new workflow requires items to be carried up a ramp and placed on a mezzanine, the Autopicker stops being an option. A humanoid, in theory, just walks over and starts the new task.

Agility calls this the “versatility premium,” although they use more polished language. The press release emphasizes that Digit is a “general-purpose, human-centric robot” and that Agility’s “embodied AI systems” enable the robot to learn new skills through deployment data, creating a “compounding advantage” over time. The 65,000 operational hours across nine customer facilities are explicitly framed as a proprietary data flywheel that continuously expands what Digit can do.

That data flywheel argument is the load-bearing wall of the $2.5 billion valuation. Strip it away and you have a hardware company with a 0.12× backlog-to-valuation ratio ($300 million in contracted orders against a $2.5 billion pre-money equity value). Defense and aerospace companies, which are the closest valuation analogues for hardware platforms with long development cycles and government-adjacent customers, typically trade at 1 to 3 times backlog. Industrial automation companies trade at 0.5 to 1 times revenue. Agility is priced like a software company.

RoboFab and the Capacity Bet

In Salem, Oregon, Agility has built RoboFab, described in the press release as a “full-scale humanoid manufacturing facility designed to support production of up to 10,000 units annually.” The company has sold roughly 100 Digit robots to date. That is one percent utilization of installed manufacturing capacity.

If Agility could fill RoboFab to capacity at $250,000 per unit, a price point that earlier Digit iterations were estimated at by the Statesman Journal, that is $2.5 billion in potential annual revenue, which makes the valuation look almost sensible at 1× forward revenue. The problem is that nobody has disclosed whether the $300 million order book represents 1,200 units at $250,000 each or 300 units with bundled multi-year software subscriptions, and the press release conditions those orders on “the realization of certain contractual milestones” that are not specified.

History offers a cautionary reference point here. SPACs have previously brought pre-revenue or early-revenue robotics companies to market with valuations anchored to total addressable market projections rather than current financials. Michael Klein’s Churchill Capital franchise has done eleven of these vehicles. CCXI shares traded at $10.54 before the announcement, barely above the $10 trust value, which tells you the SPAC market itself was pricing in zero premium for whatever target Klein would find.

What the Competition Is Actually Doing

While Agility prepares its S-4 filing, three parallel developments are squeezing the humanoid value proposition from different angles.

Tesla’s Optimus robots are operating inside Tesla factories, but Morgan Stanley reported that in early 2026, those robots were generating training data rather than performing productive labor. Elon Musk has talked about one million units in Fremont and ten million in Texas, but the Optimus 3 is not expected in high-volume production until 2027 at the earliest, and low-volume production might begin this summer, according to WCCFTech. If Tesla solves humanoid manufacturing at automotive scale, the $250,000 price point for a Digit becomes a ceiling, not a floor.

Meanwhile, Robot.com (formerly Kiwibot) has pivoted from campus delivery to wheeled humanoids with its R-noid platform, which Business Insider first detailed on June 22. Fewer than 40 units operate commercially across about a dozen sites, but the approach is telling: dual seven-degree-of-freedom arms on a wheeled mobile base, 11 pounds per arm, running on generative vision-language-action models. CEO Felipe Chavez framed it as a logical extension from delivery robots that already work, into manipulation robots that address higher-value tasks. That philosophy, start cheap and add capability upward, is the opposite of Agility’s approach, which starts at frontier robotics and tries to drive cost downward.

And NVIDIA, one of Agility’s own investors, just announced NVIDIA Halos for Robotics, the first full-stack safety architecture for physical AI. Agility is the launch partner, but the platform is explicitly designed to be hardware-agnostic, which means every humanoid competitor gains access to the same safety certification pathway that Digit v5’s “cooperative safety” story depends on.

The One Number That Would Change Everything

There is exactly one metric that could validate the versatility premium: the number of distinct tasks a single deployed Digit performs per shift. If a Digit at GXO moves totes for four hours, then walks to the loading dock and sorts packages for two hours, then navigates to a returns station and inspects items for the remaining two hours, the cost-per-tote calculation becomes irrelevant. You are no longer paying for tote-moving. You are paying for a flexible labor unit that replaces three different purpose-built systems, three different integration projects, and three different maintenance contracts.

Agility has never published this number. The 65,000 operational hours are reported in aggregate. No customer case study describes multi-task deployment in a single facility. Until that data exists, the versatility premium is a theoretical construct supporting a $2.5 billion valuation.

Limitations

Our cost-per-tote analysis has significant constraints that deserve explicit accounting. We used the $250,000 unit price estimate from the Statesman Journal because Agility does not disclose Digit pricing and that figure reflects earlier iterations, not the upcoming v5 model. The 60-moves-per-hour throughput comes from the Siemens trial of a different robot (the wheeled HMND 01 Alpha, not the bipedal Digit), and applying it to Digit may overstate the bipedal platform’s throughput by an unknown margin. Maintenance costs are estimated at 15% of capital expenditure, a standard industrial robotics assumption, but humanoid maintenance patterns at commercial scale are largely undocumented. The Brightpick comparison uses pick rates rather than tote-move rates, which are related but not identical tasks. Warehouse workers’ tote-move throughput of 45 per hour is an estimate; actual rates vary by facility layout, tote weight, and worker experience.

Strongest Counterargument

The strongest case for Agility’s valuation is not that the per-task economics work today. They don’t. The strongest case is that humanoid robotics follows the same adoption curve as industrial robots did in the 1980s, when SCARA arms cost $100,000 in 1984 dollars, performed a single task poorly, and were purchased anyway by automotive manufacturers who understood that the learning curve would eventually bend costs down while expanding capability upward. Today, a SCARA arm costs under $10,000 in real terms and performs dozens of precision tasks. Agility’s 65,000 hours of deployment data, modest as they are, represent exactly the kind of early operational experience that preceded the industrial robot cost collapse of the 1990s, and the institutional investors backing the SPAC, including NVIDIA, Amazon, SoftBank, and Foxconn, are not betting on today’s economics but on the shape of the curve. If the cost-per-task for humanoids drops by 10× over ten years (roughly matching the SCARA trajectory), the math flips entirely: a $25,000 humanoid at 60 moves per hour undercuts human labor and approaches purpose-built robot territory while retaining the ability to do twenty other things. Whether that decline actually materializes depends on manufacturing scale, AI capability improvements, and battery technology, none of which Agility controls alone, and all of which require the kind of patient capital that SPAC investors are historically not famous for providing.

What You Can Do

If you run warehouse or logistics operations: get a Brightpick, Locus Robotics, or AutoStore quote before considering a humanoid deployment. At $1,900 per month with 70-80 picks per hour around the clock, purpose-built mobile robots deliver 15 to 30 times better cost-per-task economics for single-function workflows. Only explore humanoids if your facility has three or more workflow types that change quarterly and your labor shortage is acute enough that the $126,000 annual cost of a Digit is cheaper than leaving positions unfilled, which at $22 per hour with 40% overtime premiums, means sustained vacancy rates above 25%.

If you are evaluating the AGLT SPAC: demand three numbers that the current press release does not contain: average tasks per robot per shift (the versatility proof), realized unit pricing for Digit v5 (the economics proof), and quarterly revenue or bookings cadence (the growth proof). Without those, the $2.5 billion valuation is priced on narrative, and SPAC narrative premiums historically compress 40-60% in the first twelve months of trading. Ask for the investor presentation filed with the SEC and look for the revenue bridge between $300 million in contracted orders and the $2.5 billion revenue potential at full RoboFab capacity. If that bridge relies on customers who have not yet signed, the order book is aspirational rather than contractual.

If you are building humanoid robots: publish your throughput data. The single biggest obstacle to commercial humanoid adoption is not technology. It is the absence of standardized, independently verified performance benchmarks that procurement teams can compare against existing automation. The Siemens trial is the gold standard right now, and it tested one robot doing one task for two weeks. Commission a six-month, multi-task, third-party-audited deployment study, publish the results, and watch how fast warehouse operators stop asking “do humanoids work?” and start asking “which one?”

If you are watching from the sidelines: watch the AGLT share price relative to the $10 trust value over the first 180 days after the lock-up expires. SPAC redemption rates and post-merger trading patterns will tell you whether public market investors believe the versatility premium or whether they treat this as another pre-revenue hardware story priced for a future that may take longer than the capital can sustain.

The Bottom Line

Agility Robotics has done something real. Digit works. It moves totes at GXO, operates alongside humans at Toyota and Schaeffler, and has accumulated 65,000 hours of operational data that no other humanoid company can match in commercial settings. The decision to go public through a SPAC with $620 million in fresh capital is a rational move for a company that needs to scale manufacturing from 100 units to thousands while developing the next-generation platform that might, eventually, justify a versatility premium over robots that cost a fraction as much per task.

But the $2.5 billion valuation is not paying for what Digit does today. It is paying for what humanoid robots might do in five to ten years, priced through a SPAC structure that historically discounts growth stories within months of listing, backed by institutional investors whose patience is measured in fund cycles rather than technology curves, and benchmarked against a $1 trillion addressable market that currently buys $22,800-per-year purpose-built alternatives instead. Somewhere between the three-and-a-half-cent pick and the fifty-three-cent tote move, there is a price point where versatility earns its premium. Nobody, including Agility, has yet proved where that point is, and a public listing is an expensive place to run that experiment.