Name: Apptronik Apollo
Brand: Apptronik

Who's exposed

Jobs in the threat radius

automotive assembly worker
parts kitter
production line inspector
logistics tote handler

Deployment status

Re-verified 2 August 2026, with corrections to both the generation and the wording. Apptronik unveiled Apollo 2 on 1 July 2026 in bipedal and wheeled-base configurations, and **describes it explicitly as a data collection and training platform rather than a product for sale**; the commercial machine is a third generation, Apollo 3, still in development. The previous version of this entry said 'commercially deployed,' which overstates it. What is true: Apollo units operate at Mercedes-Benz (Berlin-Marienfelde), in GXO Logistics warehouses, and at Jabil facilities, with Google DeepMind as an AI research partner — and **Apptronik has disclosed no unit count at any customer site, with no pilot publicly confirmed as converted to production.** GXO's chief automation officer, speaking about the humanoid program broadly, said 'we're not at wide-scale deployment and commercial viability yet, but we're not 10 years away,' with each robot performing a single specific task. The customer this entry previously omitted is Jabil, and it is the most informative one: under an agreement announced 25 February 2025, Jabil both **manufactures** Apollo and runs it in its own plants for inspection, sorting, kitting, lineside delivery, fixture placement and sub-assembly — Apollo building Apollo. On facilities, Apptronik expanded Robot Park in Austin, Texas to nearly 90,000 square feet in July 2026 as its flagship data collection and training site, where Apollo 2 has been the primary platform for over a year. On funding, a $520 million Series A extension closed 11 February 2026, taking the round past $935 million at a valuation above $5.5 billion, with B Capital, Google, Mercedes-Benz and PEAK6 returning and AT&T Ventures, John Deere and the Qatar Investment Authority joining. No deployment has ended.

When this hits the labor market

The previous version said one to two years for the specific Mercedes use cases and two to four years for broader light industrial deployment. That needs pushing out one notch — not because demand weakened but because the generation changed: what is running in the field is Apollo 2, a training platform, and the commercial machine, Apollo 3, does not exist yet. Until Apollo 3 launches and a checkable purchase follows, 'one to two years' has nothing to attach to. Revised read: two to three years before small-batch production work appears at existing sites like Mercedes, GXO and Jabil; four to six years before broader light industrial rollout. There is exactly one leading indicator, and it is easy to watch — **the first time any customer publishes a unit count on site.** Until that number exists, every deployment is still at the one-robot-one-task stage.

The third Western automotive humanoid — and a generation that needs correcting

If Figure goes to BMW and Atlas goes to Hyundai, Apollo goes to Mercedes. Apptronik’s $5.5B-valued bipedal humanoid works at Mercedes-Benz’s Berlin-Marienfelde facility doing exactly the tasks that drive the automotive humanoid investment thesis: bringing parts to assembly stations, inspecting components in-process, moving kitted totes between operations.

But this re-check surfaced something that has to be corrected. The Apollo in the field is now a second generation, and Apollo 2 is not a product for sale. When Apptronik unveiled it on 1 July 2026, it positioned Apollo 2 as a data collection and training platform; the commercial unit is Apollo 3, still in development. The previous version of this entry said “commercially deployed,” which was too strong, and we have replaced it with what is actually verifiable.

The customer list is one name longer

The previous version listed Mercedes and GXO. It missed Jabil — which is the most informative of the three. Under an agreement announced 25 February 2025, Jabil both manufactures Apollo for Apptronik and deploys Apollo in its own plants for inspection, sorting, kitting, lineside delivery, fixture placement and sub-assembly. Apollo building Apollo. It is a rare case in this field of the maker and the user being the same company.

The boundary of what is known also needs stating. Apptronik has disclosed no unit count at any customer site, and no pilot has been publicly confirmed as converted to production. GXO’s chief automation officer put it this way: “We’re not at wide-scale deployment and commercial viability yet, but we’re not 10 years away” — and noted each robot is doing one specific task. That sentence is closer to the truth than any launch copy.

On facilities, Apptronik expanded Robot Park in Austin to nearly 90,000 square feet in July 2026 as its flagship data collection and training site. Apollo 2 has been the workhorse there for more than a year.

One spec that was wrong

The previous spec table listed a 22-hour battery runtime. That is not right. Apollo runs roughly four hours on a charge; the 22 hours is daily operating time achieved through five-minute hot-swaps. The distinction matters: 22 hours is the output of a swap process, not the endurance of a battery.

Note also that the 173 cm, 73 kg, 71 DOF and 25 kg arm payload in this table are all Apollo 1 figures. Apptronik has published no quantitative specifications for Apollo 2 at all — its own Apollo 2 page is qualitative. The only new hard detail is configuration: Apollo 2 comes in bipedal and wheeled-base versions, with the wheeled variant aligned to industrial safety standards.

Funding is unchanged: $520 million on 11 February 2026, taking the round past $935 million at a valuation above $5.5 billion. Worth adding are the new investors — AT&T Ventures, John Deere, and the Qatar Investment Authority.

Why we care for LostJobs

The previous version argued that Apollo’s differentiator was uptime: a robot running 22 hours a day takes work that required two or three shift workers, so the labor math closes faster. The argument is sound, but this year it needs a precondition attached. That 22 hours is a four-hour battery plus a swap process, and the math has not been validated by a unit count at any customer site.

The more useful move for readers is to shift attention from what the robot can do to what actually marks progress. Across everything currently known, one indicator separates pilot from replacement: whether any customer publishes how many robots are on site. One robot doing one task is a pilot. Ten robots covering three stations on a line is replacement. The day that number appears, this entry’s timeline gets rewritten.

If you work in the Mercedes, Daimler or Bosch supply chain in Germany or the US, in a GXO-operated warehouse, or in a Jabil plant, Apollo is already in your supply chain — but its current role is apprentice, not replacement. What to watch is when Apollo 3 launches, and whether it arrives with a checkable customer order attached.