Bessemer Venture Partners presents six key predictions for the robotics and physical AI market in 2026, asserting that the industry is at a pivotal 'GPT-2.5 moment' with immense growth potential. The firm highlights the critical role of capital in building data moats, the concentration of talent leading to winner-takes-most dynamics, and the near-term value creation by full-stack, vertically integrated players. Defense robotics is identified as a leading sector for significant public market outcomes, and Bessemer argues that the overall market is currently underinvested relative to its addressable opportunity.
Every atomic assertion extracted from the underlying record, ranked by evidence strength.
The path to real-world robotics is better foundational models that understand the physical world, not better control algorithms.
The robotics market will reach $38 billion by 2035.
Real-world deployment uncovers harder, more specialized data curation and labeling problems over time.
Timelines for general-purpose robotics are a lot longer than most people expect, still five-plus years out.
Around $6B flowed into six or seven world model companies in Q1 2026 alone.
The ChatGPT moment for robotics is coming faster than most people realize.
Five years from now, the majority of robots deployed globally won't be built by today's startups, but by companies that haven't even started building robots yet.
Teleop alone will not be a successful data strategy for robotics.
The defining advantage in physical AI will be companies with the strongest data flywheel, turning robot data into better decisions, model improvements, and deployments faster than everyone else.
The defining advantage in physical AI will be the quality of the data infrastructure behind it, as models converge.
Ground robots in construction that ran $100,000 per unit a few years ago now run the same workflows for under $15,000.
Docked drones have dropped from $200,000 to under $20,000 while getting more capable.
Hardware needs to get cheap enough before deployment can scale.
The dual-use question is real and not going away, with interesting companies building systems capable enough for defense requirements that are also transformative in commercial contexts.
There will be 100,000x more robots on Earth in the next 10-20 years.
When the ChatGPT moment hits, the bottleneck will be production hours, real robots, real work, and real environments.
Companies that optimize for deployment over demos will separate decisively from the field.
Robotics is in its GPT-2.5 moment, where capabilities are real, but the gap between lab performance and field deployment remains wide.
Talent concentration will crown winners quickly in robotics, as this is not a market where 50 companies will win.
Near-term value will accrue to full-stack, vertically integrated players, not pure-play foundation model companies, because robotics foundation models aren't yet general enough to work out of the box.
Defense robotics will produce the first $50B+ IPOs in the category, driven by geopolitical tailwinds, national security urgency, and government-backed capital.
There will be no robotics bubble, and not enough capital is flowing into the industry.
Teams with the deepest expertise in sim-to-real transfer, manipulation, locomotion, and sensor fusion are building advantages that open-source releases don't easily replicate.
Team composition is the first investment decision in robotics, and it compounds faster here than almost anywhere else.
Robotics is not a market where a long tail of moderately capable companies finds niches and survives.
Deployment in robotics requires domain-specific data collection, fine-tuning for the target environment, hardware integration, and operational infrastructure to manage robots in the field.
The moats being built today in robotics aren't primarily in model architecture, but in proprietary data pipelines, domain expertise, deployment infrastructure, and customer relationships.
A warehouse automation company that has deployed hundreds of robots across dozens of facilities has millions of hours of task-specific production data.
Hardware commoditization is accelerating the dynamic of value accruing to full-stack players across verticals.
Value is concentrating in application layers today because the infrastructure layer isn't yet general enough to support end-to-end deployment on its own.
The API moment for robotics will arrive as foundation models improve and sim-to-real transfer matures.
Horizontal foundation model companies are making a long-term bet that's likely a 2028-and-beyond story.
Verticalization is where durable value is being created in the current window.
Defense robotics valuations are already pulling ahead of non-defense peers.
Median Series A post-money valuation for defense robotics companies reached $105M in 2025, compared to $50M for non-defense.
Anduril closed at a $60 billion valuation in March 2026.
Saronic raised a $1.75 billion Series D for autonomous shipbuilding in March 2026.
Defense procurement cycles are predictable, contracts are large, renewal rates are high, and switching costs are significant.
Defense buyers operate under a different calculus where capability gaps carry national security consequences, and the cost of inaction is measured in strategic risk, not dollars.
Robotics fundamentally changes the nature of modern warfare.
Nearly 90% of humanoid robots sold globally in 2025 were Chinese-manufactured.
Chinese AI models lag US models by approximately seven months on average.
The US government is beginning to treat robotics as a national security imperative.
The most defensible companies in this space are building autonomous platforms, perception systems, and decision-making infrastructure with genuine commercial application.
In the past five years, 745 software companies have raised more than $30M, compared to 42 robotics companies.
The underlying robotics market is 30 times larger than software spend globally.
Robotics is structurally underinvested relative to its addressable opportunity.
Most analysts project 50x industry growth over the next decade for robotics.
Bessemer Venture Partners thinks the 50x growth estimate anchors on automating existing workflows rather than accounting for entirely new categories of economic activity.
Transformative technology creates markets that didn't previously exist.
Not every humanoid robot company will find a viable path to deployment at scale.
Capital will consolidate around a small number of leaders in robotics.
The overall level of investment in robotics remains well below where it should be relative to the size of the opportunity and the pace of capability development.
Waiting for proof of the inflection is a sure-fire way to miss it.
Getting from 80% task success to 99.9% in robotics is not a linear problem.
The last 20% of task success requires fundamentally different approaches like tactile sensing, force feedback, and better sim-to-real transfer for manipulation.
World models and large vision-language-action models are expensive to run in real time.
Robotics models must generate an environment state every few milliseconds per robot, meaning each deployment effectively requires a dedicated GPU pipeline.
LLM inference costs dropped roughly 1,000x in three years.
Goldman Sachs revised its robotics market forecast upward sixfold in a single year.