Is Philon profitable in 2026?

No. As of July 2026, Philon has disclosed zero revenue and no named paying customers, per PitchBook's blank revenue field. The founder claimed 'expressed interest' from big AI companies in Oct 2025, but no contracts have been confirmed.

Could AI replace Philon's dataset business?

Yes, largely by substitution. Industry is shifting to synthetic data and sim-to-real transfer (NVIDIA Cosmos, Physical Intelligence's pi0.7, April 2026) which reduces demand for third-party real-world datasets like Philon's open-sourced SEA set.

Who is Philon's biggest competitor?

Cortex AI and Human Archive, both YC-backed peers pursuing an identical real-world robot-data thesis; Human Archive launched March 2026. Frontier labs like Physical Intelligence and Skild AI can also insource data collection themselves.

Philon is privately held, founded and run by 21-year-old Alexandros Petkos, originally from Greece, based in Palo Alto/San Francisco. Its sole disclosed investor is Project Europe, which put in roughly $233K on 2025-08-02 for equity.

Has Philon raised a Series A?

No. As of this writing, Philon's only disclosed funding is a single accelerator round of about $233K from Project Europe dated 2025-08-02; no seed or Series A round has been publicly reported.

Why did Philon pivot from robots to data?

Philon rebranded from an open-source general-purpose robot hardware company to a 'Spatial AI' dataset business via a YC Launch on 2025-11-17, with the founder describing it as a data-collection step toward eventually becoming a robotics company.

YC Fall 2025 · United States

Philon

Philon collects real-world video data of people doing tasks to train robot AI models.

Losing its protection

Assessed 27 July 2026 · Evidence as of 27 July 2026

The structural protection is eroding.

On the recordSince 27 July 2026 · Checkable from early 2027 · Every call on the record Checkpoint: Runway from the $233K raise runs out without a signed contract or new funding

The call

Philon sells data to the same companies that can make it themselves, with $233K raised, one employee, and rivals chasing the same buyers.

Philon collects real-world video and sensor data to help companies train robots. The problem is that its main buyers, like Physical Intelligence and NVIDIA, can easily collect this data themselves instead of paying Philon, and they are building their own synthetic data tools (like NVIDIA Cosmos) that reduce the need for real data altogether. With only about $233,000 raised and one employee, Philon lacks the resources to compete with well-funded rivals like Cortex AI, and it hurt itself by making its main dataset free and open to everyone. The one thing that could save it is landing a paying customer by early 2027, which would prove companies actually prefer buying data from Philon rather than making it themselves.

Why

AI pressure: NVIDIA Cosmos, Physical Intelligence's pi0.7. Cosmos and pi0.7 generate or generalize training data on their own, cutting the need to buy datasets like Philon's. Could AI do the core job? Yes, likely.

Bigger players: Physical Intelligence, Skild AI, NVIDIA. Philon's biggest buyers are also its biggest threats: they can build their own data pipelines and stop paying Philon. Could a giant copy it? Yes, easily.

What protects it. The founder's robotics background and a free, open dataset draw attention, but nothing stops labs from copying the approach themselves.

What would change this call

Philon discloses a signed, named paying customer contract by Q1 2027. Would show frontier labs choose to buy rather than build their own data pipelines.
Philon closes a priced follow-on seed or Series A before its ~$233K runway depletes. Would show investor confidence and survival past the early 2027 cash deadline.
Cortex AI or Human Archive announce a frontier-lab data contract before Philon does. Would show a same-thesis rival winning the customers Philon is targeting.

Key numbers

$233K total raised, Project Europe accelerator (2025-08-02)
1 employee listed on Prospeo (2025-08-30)
Skild AI: $1.4B raised, ~$30M ARR reported (2026-04-08)

Named risks

No disclosed revenue or named paying customers as of mid-2026 despite founder claims of interest. Severity: high
Only about 1 employee and $233K raised versus rivals with $400M to $1.4B in funding. Severity: high
Open-sourced SEA dataset removes exclusivity, letting buyers replicate the asset for free. Severity: high
Frontier-lab customers (NVIDIA, Physical Intelligence, Skild AI) can insource data collection they'd otherwise buy from Philon. Severity: medium
Company has already pivoted once from hardware to data licensing, suggesting strategic instability. Severity: medium

Rivals, as read in July 2026

Rival	Direction	Evidence
Cortex AI	contesting	YC peer with identical real-world robot-data thesis, launched around same period as Philon's pivot
Human Archive	gaining on	Launched 2026-03-08 targeting the same robot-data bottleneck, newer and potentially better resourced
Physical Intelligence	losing to	Shipped pi0.7 model (2026-04-16) generalizing to new tasks, cutting reliance on third-party datasets like Philon's
Skild AI	losing to	Raised $1.4B and reached ~$30M ARR by 2026-04-08 versus Philon's $233K raised and no disclosed revenue
NVIDIA	losing to	GR00T N2 (March 2026 GTC) reportedly doubled task-success rates, reducing demand for external real-world data

Questions people ask

Is Philon profitable in 2026?: No. As of July 2026, Philon has disclosed zero revenue and no named paying customers, per PitchBook's blank revenue field. The founder claimed 'expressed interest' from big AI companies in Oct 2025, but no contracts have been confirmed.
Could AI replace Philon's dataset business?: Yes, largely by substitution. Industry is shifting to synthetic data and sim-to-real transfer (NVIDIA Cosmos, Physical Intelligence's pi0.7, April 2026) which reduces demand for third-party real-world datasets like Philon's open-sourced SEA set.
Who is Philon's biggest competitor?: Cortex AI and Human Archive, both YC-backed peers pursuing an identical real-world robot-data thesis; Human Archive launched March 2026. Frontier labs like Physical Intelligence and Skild AI can also insource data collection themselves.
Who owns Philon?: Philon is privately held, founded and run by 21-year-old Alexandros Petkos, originally from Greece, based in Palo Alto/San Francisco. Its sole disclosed investor is Project Europe, which put in roughly $233K on 2025-08-02 for equity.
Has Philon raised a Series A?: No. As of this writing, Philon's only disclosed funding is a single accelerator round of about $233K from Project Europe dated 2025-08-02; no seed or Series A round has been publicly reported.
Why did Philon pivot from robots to data?: Philon rebranded from an open-source general-purpose robot hardware company to a 'Spatial AI' dataset business via a YC Launch on 2025-11-17, with the founder describing it as a data-collection step toward eventually becoming a robotics company.

Timeline

2026-07-06Unrelated German firm PHILON Service Robotics AG announces FieldBots partnership, causing brand confusion
2026-04-16Physical Intelligence unveils pi0.7 model generalizing to untrained tasks, reducing third-party data need
2026-04-08Skild AI reported at $1.4B raised and ~$30M ARR; NVIDIA GR00T N2 doubles task-success rates
2026-03-08YC peer Human Archive launches, competing directly on same robot-data thesis
2026-01-26CB Insights notes industry shift toward synthetic data over third-party datasets
2025-11-17Philon relaunches on YC as 'Spatial AI', pivoting from robot hardware to data licensing
2025-11K-Scale Labs, similar YC/Palo Alto open-source robotics startup, shuts down after failed Series A
2025-10-03Founder Alexandros Petkos describes plan to monetize dataset by year-end 2025
2025-08-30Prospeo lists Philon headcount at 1 employee
2025-08-02Philon raises $233K accelerator round from Project Europe

Revision history

27 July 2026

First assessed. Losing its protection. The text above is fixed as of this date and is not edited afterwards.

In the same band

Nexus Terrain Metorial