Andra Keay's June robotics ledger turns funding into physical AI's scorecard

Why it matters

Robotics funding is becoming the market's proxy for physical AI momentum, but June's $6.57 billion tally also shows how disclosure gaps and defense valuations can blur real traction.

Andra Keay put a market number on the physical AI race on July 5: her Robots & Startups funding roundup counted 136 robotics funding rounds in June 2026 totaling $6,570,460,729.

Keay is a useful messenger for this particular moment because she has spent years close to robotics founders rather than treating robots as a side category of AI. Her Robots & Startups newsletter says it covers robot startups from Silicon Valley Robotics, and her public bio identifies her as managing director and founder of Silicon Valley Robotics, the industry group around Bay Area robotics commercialization.

That matters because the June tally is less a clean leaderboard than a temperature reading. Keay's own caveats do the important work: roughly 33% of entries had no reported funding amount, and large late-stage rounds skewed the monthly total. In other words, the $6.57 billion figure says investors are crowding into robotics and embodied AI, while saying less about which companies have durable revenue, defensible deployment data, or hardware economics that survive outside a pitch deck.

The reason the number still travels is simple. Software AI has trained venture investors to look for model capability, usage growth, and enterprise adoption curves. Physical AI gives them a messier scorecard: contracts, factories, component supply, field reliability, regulatory clearance, defense procurement, autonomy performance, and the number founders can disclose most easily, capital raised.

Defense is pulling the category toward bigger checks

Keay opened the June roundup with the category that has made the robotics funding curve harder to read: defense. She cited a Fortune piece in which Anduril co-founder and CEO Brian Schimpf said the defense-tech market has entered bubble territory in some corners. Fortune reported that VCs deployed $19.8 billion into defense tech across 262 deals in the first quarter of 2026, citing PitchBook, up from $5.7 billion in Q1 2024 and about $17 billion in Q1 2025.

Schimpf has standing to make that warning because Anduril is one of the companies that created the market investors are now chasing. In May, Schimpf wrote in an Anduril announcement that the company had raised a $5 billion Series H at a $61 billion valuation, led by Thrive Capital and Andreessen Horowitz. In the same note, he said that when Anduril was founded in 2017, defense was not a category that attracted significant venture investment.

That reversal is now the context for every robotics founder raising money. A startup building autonomy, drones, industrial robots, simulation, sensing, or manufacturing infrastructure can be read through two frames at once: a robotics company trying to ship machines, and a physical AI company trying to claim strategic relevance. Defense gives that second frame budget urgency and buyer seriousness. It also invites inflated comparisons.

Large defense programs can give investors something rare in frontier robotics: a path from prototype to government-backed production. Most robotics founders raising in the same market do not have that signal. Their rounds may still get priced against a market that has learned to use Anduril as the reference point.

The June list shows breadth, not proof

The visible portion of Keay's June list names Bay Area companies that raised during the month, including Generalist AI, Rokid, Lightwheel, Tombot, Dexterity, Proception, Nuro, Queue, Uvify, Applied Intuition, InLoop Robotics, Ornadyne, Intelligence Factory, Complete Robot, Andustry, Advanced Metal Research, Arlo Industries, and TwoLab.

That spread is the story. The money is not landing in one robotics lane. It is flowing across autonomous vehicles, industrial automation, AI systems for the physical world, robotics hardware, simulation, and likely dual-use categories. Some of those companies are closer to software margins. Others are closer to manufacturing reality, where the cash needs are larger and the iteration cycles are slower.

The category label can hide that difference. A physical AI startup with a large simulation or developer-tool component can scale differently from a company that must build, install, service, and replace machines in the field. A defense-adjacent autonomy company may have government demand tailwinds, while a consumer robot has to survive price sensitivity, support costs, and distribution. Both can show up in a robotics funding roundup. They do not carry the same risk.

That is why Keay's disclosure gap matters. If one-third of entries lack reported amounts, the monthly total undercounts some activity while overemphasizing the rounds that disclose large numbers. Late-stage financings can make a month look like a boom even if early-stage formation is thinner, or if follow-on capital is concentrating around a small set of perceived winners.

Founders now raise against a harder benchmark

For robotics founders, the market is friendlier than it was during the long period when hardware risk scared generalist venture firms away. The check writers who once preferred pure software now have a language for autonomy, embodied AI, and defense-industrial urgency. That can help serious teams fund the unglamorous work that robotics requires: manufacturing, safety testing, reliability engineering, supply chains, fleet operations, and customer support.

The tradeoff is that the benchmark has moved. If investors read robotics funding as the scorecard for physical AI, founders are pushed to explain why their system deserves AI multiples while carrying hardware obligations. A high valuation can buy runway and manufacturing capacity. It can also lock a company into expectations that assume deployment curves will look more like software than machines.

Schimpf's warning lands here. Anduril can argue for its valuation with defense contracts, revenue scale, manufacturing investment, and a software layer in Lattice that ties systems together. Even then, the company is the exception investors cite when they want to believe a new defense or robotics company can become a platform. The risk is that weaker companies borrow the same story without the same procurement path, revenue base, or production discipline.

Keay's June roundup should be read as evidence of capital intent. Investors want exposure to the next wave of embodied intelligence. They are funding the companies that might turn AI from a screen-based product into machines that move through warehouses, roads, factories, homes, and battlefields. The funding totals show where the market wants the future to be.

They do not prove the future has arrived. The proof will be in deployed fleets, repeat buyers, working unit economics, field uptime, safety records, and contracts that survive beyond pilots. Until those metrics are as visible as the rounds, robotics funding will remain the easiest scorecard for physical AI and one of the noisiest.