PORTFOLIO

Portfolio Update: Fireside Chat w/ Gary Chen of Raise Robotics

Topics covered: where robotics stands today, hype vs. reality, and key breakthroughs driving progress. We touched on which industries are likely to be transformed first, the biggest bottlenecks ahead, how Al is accelerating robotics, and where early investors might want to place their bets.

Company Snapshot

Founder & CEO: Gary Chen
Background: Waymo + Cruise (autonomous vehicle simulation, perception, planning).
Raise Robotics builds mobile manipulators for construction and heavy manufacturing.
Raised ~$10M to date, 16 units in production, deployed on real projects.
Core value: automate high-precision, hazardous, hard-to-reach tasks (e.g., welding at heights, storage tanks, ceilings).

Key Learnings from the Discussion

1. Robotics Today (Hype vs. Reality)

Reality = robots generating revenue now in niche verticals (logistics, construction, agtech, lab automation).
Hype = humanoids/general-purpose robots — still a 10+ year horizon before broad deployment.
Gary: “Revenue is reality, hype is everything else.”

2. Industry Roadmap

2 years: Scaling existing deployments (warehouses, construction automation, lab robotics).
5 years: Autonomous vehicles becoming mainstream in major U.S. cities, more vertical-specific robotics maturity.
10 years: Humanoids/fully general-purpose robotics start scaling — but not viable today.

3. U.S. vs. China vs. Europe

China is pulling ahead in industrial robotics adoption and manufacturing culture.
Manufacturers there take startups seriously regardless of size; in the U.S./Europe, startups often struggle to get attention.
Cultural difference: China prioritizes near/mid-term improvements; West over-invests in long-term bets with slower payoffs.

4. Edge Cases Dominate Robotics

Robotics = constant edge cases.
Unlike software, edge cases can take years to fix and often require re-architecting entire systems.
Many startups fail because they underestimate edge-case cost and burn through capital.

5. Open Source & Foundation Models

Open source accelerates early-stage robotics development, but domain-specific data remains critical.
Skeptical that a “foundation model for robotics” will generalize well across industries.

6. Man–Machine Interfaces

Customers care about time to integration (how fast a robot can be dropped into workflows).
Ideal future: natural language/“tell the robot what to do.”
Hard truth: the biggest barrier isn’t capability, it’s testability — proving reliability across tasks at scale.

7. Trade-offs in Building Raise

Didn’t have hundreds of millions to burn → forced to design robots with cost in mind from day one.
Chose to hit BOM (bill of materials) targets early rather than prototype with expensive components and cut costs later.
Belief: startups that design with realistic costs from the start are more likely to succeed.

8. Simulation

Heavy user of simulation tools for development, learned best practices at Waymo.
Not perfect, but accelerates iteration and cuts down expensive real-world testing.

9. Humanoids vs. Task-Specific Robots

Short term: Customers want experts (robots solving one problem extremely well).
Generalists/humanoids: won’t be adopted until they can deliver guaranteed reliability.
Example: A humanoid robot can’t lift a 40-ton steel tank or hit micron-level precision in welding.
Humanoids better suited for logistics, bin-picking, or human-task substitution — not heavy industry.

10. Lifespan & Maintenance

Industrial robotics customers expect 5–10+ year hardware lifespans; some still maintain 30–40 year old robots.
Consumer robots will likely follow the iPhone model (planned obsolescence every 3–5 years).
Raises questions about distribution + maintenance as the bottleneck, not technology.

Overall Takeaways

Adoption is vertical-specific — each industry will have its own “ChatGPT moment.”
Revenue beats hype. If the robot isn’t generating ROI today, it’s just an R&D project.
China is leading on manufacturing culture and willingness to partner with startups.
The biggest challenge isn’t building capability — it’s managing edge cases, testing, cost discipline, and customer integration.