Fern
Two-person startup building RL training environments for robotics AI models.
On the record
Checkpoint: Whether Fern stays in robotics or fully shifts to its meeting-AI copilot becomes clear
The call
Fern is a tiny team pitching robotics training tools into a space that four $1B-plus labs and NVIDIA already occupy, and it already launched a second product in a different market.
Fern is a two-person startup trying to create virtual environments and AI world-models that teach robots how to move and act. The problem is that bigger companies like Physical Intelligence, World Labs, and AMI Labs are already building almost identical products with billions of dollars, while NVIDIA (which makes the chips everything runs on) is also building free tools that do the same thing. The founders themselves started moving toward meeting software in early 2026, which suggests they concluded the robot space is too crowded and expensive to win.
Why
01
AI pressure: World Labs (Marble), Physical Intelligence, AMI Labs, Skild AI. These labs raise $1B+ each to build their own world models and training data, the exact layer Fern is trying to sell into. Could AI do the core job? Yes, likely.
02
Bigger players: NVIDIA (Isaac Lab, Isaac Sim, Cosmos). NVIDIA already ships free, widely used robotics simulation and training tools that Fern's environments would have to compete against or sit on top of. Could a giant copy it? Yes, easily.
03
What protects it. None apparent, no exclusive data deals or locked-in customers have surfaced yet.
What would change this call
Fern announces a Series A or additional funding round within 12 months. Would show investors still believe in the robotics thesis despite the pivot.
Fern fully drops the meeting-AI copilot or fully drops robotics RL product. Would confirm the founders chose one viable path over an unwinnable one.
Fern announces a named exclusive data or design-partner deal with a robotics fleet, factory, or hospital. Would show Fern secured a defensible niche the $1B+ labs cannot easily match.
Fern fully drops the meeting-AI copilot or fully drops robotics RL product. Would confirm the founders chose one viable path over an unwinnable one.
Fern announces a named exclusive data or design-partner deal with a robotics fleet, factory, or hospital. Would show Fern secured a defensible niche the $1B+ labs cannot easily match.
Key numbers
- Fern has raised $500K total on a convertible note
- Team is 2 people versus Skild AI's $1.4B raise
- Physical Intelligence in talks at ~$11B valuation
Named risks
- Two-person, $500K-funded team competing against four $1B+ funded robotics rivals.
- Unclear or dual product identity between robotics RL infra and meeting-AI copilot.
- No verified customers, revenue, or data partnerships in any third-party source.
- Brand confusion with two unrelated companies also named Fern.
Rivals, as read in July 2026
| Rival | Direction | Evidence |
|---|---|---|
| Physical Intelligence | losing to | Talks to reach ~$11B valuation by 2026-04-16 versus Fern's $500K total raise |
| Skild AI | losing to | $1.4B raise at $14B valuation with ~$30M revenue as of 2026-01-31 |
| World Labs | losing to | $1B raised 2026-02-19, Marble product explicitly targets robotics applications |
| AMI Labs | losing to | LeCun-led lab raised over $1B seed for robotics world models, 2026-03-10 |
| NVIDIA (Isaac Lab/Cosmos) | contesting | Already the dominant sim/RL training stack with no conflict-of-interest barrier |
| One Robot (YC W26) | contesting | Same YC batch, nearly identical world-models-for-robots pitch launched 2026-02-24 |
Questions people ask
- Is Fern (the YC robotics startup) profitable in 2026?
- No revenue has been verified for Fern in any third-party source as of the June 17, 2026 CB Insights snapshot. The company runs on a $500K convertible note with a two-person team, and no customer names or pricing have been disclosed, so profitability cannot be confirmed.
- Could NVIDIA replace Fern's robotics product?
- Yes, structurally. NVIDIA's Isaac Lab and Cosmos already provide the dominant simulation and RL training stack for robotics as of 2026, occupying the same neutral infrastructure layer Fern targets, without the conflict-of-interest issue Fern claims to solve.
- Who are Fern's biggest competitors?
- Physical Intelligence (near $11B valuation talks by April 2026), Skild AI ($14B valuation, ~$30M revenue, Jan 2026), World Labs ($1B raised Feb 2026), and Yann LeCun's AMI Labs (>$1B seed, March 2026) all target the same robotics world-model space Fern entered with just $500K.
- Did Fern pivot away from robotics?
- Fern launched a realtime meeting-AI copilot on February 26, 2026, but CB Insights' June 17, 2026 batch report still described Fern as a robotics RL-environment company, leaving its actual product focus unresolved as of mid-2026.
- Who owns Fern?
- Fern is privately founder-owned, originally incorporated as Ishiki Labs in 2025 by ex-Meta LLaMA and Reality Labs researchers Amit Yadav and Robert Xu. It has raised $500K on a convertible note as of June 2026, with no institutional lead investor identified.
- How much funding has Fern raised?
- Fern has raised a total of $500K via convertible note, per CB Insights data referencing a raise roughly six months before its June 2026 snapshot. This compares to $1B or more raised by each of its four main robotics competitors in the same period.
Timeline
- 2026-06-17CB Insights still classifies Fern as a robotics/physical-AI infrastructure company
- 2026-04-16Physical Intelligence ships pi0.7 model, valuation talks near $11B
- 2026-03-10Yann LeCun's AMI Labs raises more than $1B seed round for world models
- 2026-02-26Fern launches realtime meeting-AI copilot, pivoting from robotics pitch
- 2026-02-24YC batchmate One Robot launches near-identical world-models-for-robots pitch
- 2026-02-19Fei-Fei Li's World Labs raises $1B for Marble spatial world model
- 2026-01-31Physical Intelligence raises $1B; Skild AI reported at $1.4B raise, $14B valuation
Revision history
27 July 2026
First assessed. Little protection left. The text above is fixed as of this date and is not edited afterwards.
In the same band
Sources
- Fern: RL environments for robotics companies
- Products, Competitors, Financials, Employees, Headquarters Locations
- Jobs at Fern (Formerly Ishiki Labs)
- Launch YC: Fern, Realtime Socially Aware AI for Meetings (2026-02-19)
- Fern Launches: Realtime Socially Aware AI for Meetings (2026-02-26)
- Physical Intelligence... new robot brain can figure out tasks it was never taught (2026-04-16)
- Physical Intelligence raises $1B to build robot brains, no revenue plan yet (2026-01-31)
- Fei-Fei Li's World Labs raises £1bn to advance spatial intelligence (2026-02-19)
- Yann LeCun's AMI Labs Raises $1B Seed Round (2026-03-10)
- Launch YC: One Robot (YC W26) - World models for robots (2026-02-24)
- How Isaac Lab Accelerates Reinforcement Learning
- Y Combinator's Winter 2026 batch is its most technically complex cohort yet (2026-06-17)