Apptronik

Apptronik is an American humanoid-robotics company headquartered in Austin, Texas, founded in 2016 by Jeff Cardenas, Nick Paine, and Luis Sentis as a commercial spinout from the University of Texas at Austin Human Centered Robotics Lab. Apptronik develops the Apollo bipedal humanoid robot, a five-foot-eight-inch electric humanoid designed for warehouse, manufacturing, and logistics deployment, alongside a portfolio of prior research humanoid platforms built under prior NASA-collaboration programs. As of 2026, Apptronik is one of the principal American humanoid-robotics insurgents alongside Figure AI, 1X, Agility Robotics, and Tesla AI's Optimus program, with a Mercedes-Benz factory pilot, a long-running NASA partnership, and approximately $400 million in cumulative private capital raised through early 2026.

At a glance

• Founded: 2016 in Austin, Texas by Jeff Cardenas, Nick Paine, and Luis Sentis.
• Status: Private.
• Funding: Approximately $400 million cumulative. $350 million Series A in March 2024 co-led by Capital Factory and B Capital Group; Series B in February 2025 led by Google (Alphabet) with Ark Invest, Mercedes-Benz, and Sutter Hill Ventures participating. Specific Series B size publicly described as a multi-hundred-million-dollar round at a multi-billion-dollar post-money valuation.
• CEO: Jeff Cardenas, co-founder.
• Other notable leadership: Nick Paine (Chief Technology Officer, co-founder), Luis Sentis (Chief Scientific Officer, co-founder, University of Texas at Austin professor), Steve Sherman (Chief Operating Officer).
• Open weights: None. Apptronik is a humanoid-hardware company rather than an AI foundation-model company; the AI control stack is proprietary, with selected partnerships using Google DeepMind models as the AI substrate.
• Flagship product: Apollo bipedal humanoid robot (unveiled August 2023, commercial pilots running by 2024 to 2026).

Origins

Apptronik was founded in 2016 as a commercial spinout from the Human Centered Robotics Lab at the University of Texas at Austin, directed by Luis Sentis. Sentis had spent more than a decade running robotics research at UT Austin, with a focus on series-elastic actuators, whole-body control, and bipedal locomotion. The lab had been a long-running NASA collaboration partner on the Valkyrie humanoid robot research program, the agency's flagship humanoid platform developed from 2013 onward for potential extraterrestrial and disaster-response use cases.

The founding decision in 2016 was driven by an observation that the bipedal-humanoid commercial market was not yet open (Boston Dynamics' Atlas was a research platform, the Honda ASIMO program had wound down, no commercial humanoid was deployed at scale anywhere) but that several enabling technologies (electric series-elastic actuators, lithium-ion batteries with appropriate power density, embedded compute for real-time control, computer-vision foundation models) were trending toward commercial viability. Apptronik's founding thesis was that a company building the next-generation humanoid hardware platform would be positioned to capture the market when the AI and battery technologies converged.

The first decade of the company was concentrated on hardware research and on NASA-collaboration humanoid prototype development. Apptronik built or contributed to multiple research humanoid platforms (Astra, Apollo predecessors, the DRACO 3 bipedal robot, and others) before the August 2023 Apollo commercial reveal positioned the company on the public humanoid commercial map. The Apollo design borrows heavily from the lessons of those research platforms: an electric drive train, a five-foot-eight-inch human-scale form factor, two-arm grasp-and-manipulation capability, and a swappable-battery architecture for continuous operation in industrial shift work.

The Mercedes-Benz factory pilot announced in March 2024 was the company's first major automotive-OEM customer; the deployment at the Mercedes-Benz factory in Hungary (and later expanded to other sites) tests Apollo in real automotive-line manufacturing tasks such as part transport, kitting, and quality inspection. The Mercedes partnership is structurally important as a proof-of-concept that humanoid robots can integrate into existing automotive-manufacturing workflows without requiring substantial line redesign, a claim that has been central to the humanoid-robotics commercial thesis since the early 2020s.

Mission and strategy

Apptronik's stated mission is to "create a future where robots are universally helpful and create equitable abundance," operationalised through Apollo as a general-purpose industrial humanoid robot. The strategic premise is that the most valuable initial humanoid deployment surface is industrial work (warehouses, manufacturing, logistics) rather than consumer applications, because industrial environments are structured enough that current AI control systems can handle them reliably, and because industrial labour shortages provide a clear customer pull at price points that justify the unit economics.

The vertical-versus-horizontal positioning is in the middle of the humanoid market. Apptronik builds the hardware itself but relies on partner foundation models for the AI control stack, most visibly through the announced partnership with Google DeepMind (revealed in late 2024). The partnership applies DeepMind's robotics-AI models to the Apollo platform, with Apptronik focused on the hardware-and-deployment layer and DeepMind focused on the perception-and-policy layer. The split-stack strategy distinguishes Apptronik from vertically-integrated competitors like Figure AI (which builds both the hardware and the Helix VLA model in-house) and Tesla AI (which builds both the Optimus hardware and the FSD-derived control stack).

The capital-efficient hardware approach is part of the strategic differentiation. Apptronik's Series A and Series B sizes are smaller than Figure AI's funding totals (approximately $400 million versus $1.5 billion through similar rounds), reflecting the company's decision to focus on hardware-and-integration efficiency rather than on simultaneously building a frontier-scale AI training operation. The bet is that the AI foundation-model layer commoditises faster than the hardware-and-deployment layer, and that companies that own the hardware-and-integration position will capture more of the long-run value than companies that try to vertically integrate everything.

Models and products

• Apollo (commercial humanoid, unveiled August 2023). Five-foot-eight-inch bipedal humanoid robot weighing approximately 160 pounds, designed for industrial deployment with payload capacity of around 55 pounds. Battery operation for approximately four hours per swap, with a swappable-battery architecture supporting continuous-shift operation. Two-arm manipulation with five-finger end-effectors. Force-controlled compliant actuators for safe human-collaborative operation. Computer-vision and language-instruction interfaces.
• Research humanoid lineage. Earlier platforms include Astra, Apollo predecessors, and the DRACO 3 (a smaller research bipedal robot). The NASA Valkyrie collaboration, which predates the company's founding, contributes substantially to the institutional engineering knowledge that underlies the Apollo design.
• AI partnership stack. No first-party AI foundation models. The control stack uses partner foundation models, most visibly through the announced Google DeepMind collaboration applying DeepMind's robotics models to Apollo. Selected internal control software (whole-body controllers, low-level actuator software) is proprietary.

Benchmarks and standing

The humanoid-robotics category has no widely-accepted standard benchmarks at the level that the language-model category has (LMArena, MMLU, SWE-bench). Standing in the humanoid category is currently measured through (1) named customer deployments, (2) hours of operating time per platform, (3) demonstrated task coverage on industrial tasks, and (4) capital raised.

By those measures, Apollo is one of the four or five most-credible commercial humanoid platforms globally as of 2026, alongside Figure AI's Figure 02, Tesla's Optimus, 1X's NEO and EVE, and Agility Robotics' Digit. The Mercedes-Benz pilot has run continuously since 2024, putting Apollo's customer-deployment evidence in the lead group; the Google DeepMind partnership is structurally meaningful for the AI-control-stack credibility. Apptronik does not publish benchmark-grade evaluation data; the standing-by-deployment-evidence framing is the current state of the category broadly.

Leadership

• Jeff Cardenas (Chief Executive Officer, co-founder). Prior to Apptronik, Cardenas was an engineer at the UT Austin Human Centered Robotics Lab and held adjacent commercial-robotics-engineering roles. The CEO role at Apptronik is his first chief-executive position; the company's growth from spinout through Series B has been substantially led by Cardenas's operator-side trajectory.
• Nick Paine (Chief Technology Officer, co-founder). UT Austin PhD in robotics, focused on series-elastic actuator design and whole-body humanoid control. Drives the engineering organisation.
• Luis Sentis (Chief Scientific Officer, co-founder). UT Austin professor of mechanical engineering and director of the Human Centered Robotics Lab. Long-running NASA collaboration history; the academic anchor for the company's research lineage.
• Steve Sherman (Chief Operating Officer). Operations executive recruited to scale the manufacturing and deployment organisation.

The senior engineering team includes named hires from former Boston Dynamics, NASA Johnson Space Center, and other senior-robotics-engineering institutions, though the public-facing roster has been smaller and more concentrated than the Figure AI or Boston Dynamics teams.

Funding and backers

Apptronik has raised approximately $400 million in cumulative private capital through early 2026.

• Series A (March 2024): $350 million co-led by Capital Factory and B Capital Group. The round was unusually large for a Series A in the humanoid-robotics category, reflecting the Mercedes-Benz partnership announcement and the perceived strategic value of an established NASA-collaboration spinout.
• Series B (February 2025): Led by Google (Alphabet) with Ark Invest, Mercedes-Benz, and Sutter Hill Ventures participating. The Google-led round formalises the broader Google DeepMind robotics-collaboration relationship at the capital level. Specific round size and post-money valuation were publicly described in the multi-hundred-million-dollar and multi-billion-dollar range respectively, with industry coverage placing the post-money in the $1.5 billion to $2 billion range.

The capital-side trajectory follows the talent-leads-capital pattern documented in the diaspora map but at the hardware-credentials tier rather than the frontier-AI-tier. Apptronik's Series A pricing was unusually compressed by the NASA-collaboration history and the Mercedes-Benz pilot, both of which are credentials that compress the time-to-credibility for institutional capital independent of any AI-foundation-model frontier credential.

Industry position

Apptronik sits in the American humanoid-robotics commercial cohort alongside Figure AI, Tesla AI's Optimus, 1X, and Agility Robotics. Within that cohort, Apptronik is positioned as the industrial-warehouse-and-manufacturing entrant with a long-running NASA research lineage and a major automotive-OEM deployment pilot.

The split-stack strategy (Apptronik does hardware; partner foundation models do AI) differentiates the company from the vertically-integrated competitors and creates a specific commercial bet: that hardware-and-deployment specialisation captures more long-run value than full vertical integration. If the bet is right, Apptronik becomes the AWS-of-humanoids (the infrastructure layer that multiple AI partners use); if the bet is wrong, the AI-and-hardware-co-design integration advantage that Figure AI and Tesla pursue produces better unit economics and Apptronik becomes a commodity hardware supplier.

The Chinese humanoid-robotics cluster (Unitree, Agibot, UBTECH, XPeng Robotics, and others) competes globally on price and on manufacturing scale, with the structural cost advantage of operating inside the Chinese hardware supply chain. Apptronik's defence against the Chinese cluster is the US-customer-and-regulatory-environment fit and the established commercial pilot relationships, but the cost-side comparison will tighten through 2026 and 2027.

Competitive landscape

• Figure AI (US): The closest direct American competitor. Vertically integrated (own hardware plus Helix VLA model). Approximately $1.5 billion raised through Series C, valued at $39.5 billion as of February 2025. Industrial deployment focus with BMW partnership.
• Tesla AI Optimus (US): Highest-volume distribution potential through Tesla's existing manufacturing infrastructure. Internal deployment in Tesla factories; external commercial launch not yet realised.
• 1X (Norway / US): Consumer-and-household positioning (NEO, EVE) rather than industrial. $375 million cumulative raised. OpenAI Startup Fund backing.
• Agility Robotics (US): Warehouse-and-logistics focused with Amazon deployment relationship. Digit platform.
• Boston Dynamics (US, owned by Hyundai): Atlas (humanoid, transitioned to all-electric in 2024) plus Spot (quadruped) and Stretch (warehouse). Strongest research-grade engineering lineage in the global humanoid market.
• Chinese humanoid cluster: Unitree, Agibot, UBTECH, XPeng Robotics, Xiaomi Robotics, Fourier, EngineAI. Cost-and-volume positioning. Different export-control and customer-procurement dynamics than US cohort.
• European entrants: Neura Robotics (Germany), Hexagon Robotics (Sweden/Switzerland). Smaller scale and different vertical orientations.

Outlook

Open questions and watchable signals over the next 6 to 18 months:

• Mercedes-Benz pilot expansion or contraction. The Mercedes pilot has been Apptronik's most-visible customer relationship. Whether it expands to additional Mercedes plants, additional automotive OEMs, or additional task domains will be a leading indicator of commercial traction.
• Apollo unit-economics disclosure. No company in the humanoid-robotics category has publicly disclosed the per-unit cost or the per-hour-of-operation cost of their commercial humanoids. The first credible disclosure from any of the leading platforms (including Apollo) will reset the public framing of the category's revenue trajectory.
• The Google DeepMind partnership at deployment scale. The partnership has been announced and demonstrated; the question is whether it translates to deployed performance gains in the Mercedes pilot and in any subsequent customer relationships. A first-public-deployment-side benchmark would materially shift the credibility of the split-stack strategy.
• Series C or later round pricing. A subsequent funding round in 2026 or 2027 will reveal the institutional-capital view of Apptronik's strategic position relative to Figure AI, Tesla Optimus, and the broader humanoid cohort. The Series B at $1.5 to $2 billion post-money was a substantial step from the Series A pricing; the next step will indicate whether the deployment-evidence is converting to revenue-trajectory belief.
• The Chinese-cluster cost-competition response. As Chinese humanoid platforms enter the market at lower price points, the question is whether Apptronik's customer relationships, regulatory-environment fit, and partner-stack quality provide enough differentiation to sustain US-market pricing. The 2027 market-share data will be the first test.

Sources

• Apptronik Apollo product page. The official platform description and capability framing.
• Apptronik raises $350 million Series A. Bloomberg coverage of the Series A round.
• Apptronik announces Mercedes-Benz partnership. The company's announcement of the automotive-OEM pilot.
• Apptronik and Google DeepMind partnership. The AI-stack partnership announcement.
• University of Texas at Austin Human Centered Robotics Lab. The academic lab that originated the company's research lineage.
• Companion profiles: Figure AI, Tesla AI, 1X, Agility Robotics, Boston Dynamics for the broader humanoid-robotics cohort context, and Google DeepMind for the AI partnership stack.