How much has Physical Intelligence raised?

Physical Intelligence has raised approximately $1.07 billion in confirmed primary equity across three rounds: a $70 million seed (March 2024), a $400 million Series A (November 2024), and a $600 million Series B (November 2025). All disclosed financing is equity — no venture debt, revenue-based financing, or convertible notes have been publicly reported. Fourteen named institutional and individual investors appear across the three rounds, including both venture funds (Thrive, Lux, Sequoia, Khosla) and crossover public-market investors (T. Rowe Price, Index Ventures).

By any standard, the pace is extraordinary. Physical Intelligence went from founding to over $1 billion raised in roughly 20 months — a cadence historically associated with frontier LLM labs like Anthropic and Mistral, not robotics startups. No venture debt has been reported, which is notable at this scale and suggests investors have provided runway without requiring debt instruments.

If the reported ~$1 billion Series C closes at an $11 billion valuation, total primary equity raised would approach $2.1 billion in approximately two years of operation. That would put Physical Intelligence ahead of many publicly traded robotics companies by capitalization, on the strength of models and commercial pilots rather than shipping hardware.

The investor roster spans the full spectrum of frontier-AI and growth investing. Thrive Capital and Lux Capital have backed every round from seed through Series B, providing the continuity that signals genuine long-term conviction. Thrive — the firm behind early bets on Stripe and GitHub — is one of the most influential late-stage software growth firms; its consistent participation anchors each round's credibility.

Jeff Bezos joined at Series A as a personal (non-Amazon) investor and returned for Series B. His participation adds strategic credibility given his deep interest in robotics and warehouse automation, and signals that he views Physical Intelligence as complementary to (or competitive with) Amazon's own robotics investments. OpenAI backed the company at seed and Series A — a notable strategic alignment given OpenAI's own ambitions in embodied AI. CapitalG's lead on Series B is the most interesting signal: it is a strategic endorsement from Alphabet, whose Google DeepMind unit is building directly competing models. Index Ventures and T. Rowe Price (a crossover public-market investor) provide late-stage balance-sheet depth and establish a cleaner path to IPO due diligence.

If Founders Fund and Lightspeed join in Series C as reported, the cap table will span virtually every tier of Silicon Valley investing — from research-stage seed (Lux) through growth (CapitalG, Index) to buyout-adjacent (T. Rowe Price). That breadth is unusual for a two-year-old company and reflects how consensus has formed around robotics foundation models as a strategic category.

Physical Intelligence's valuation — from roughly $400M post-seed to a reported $11B Series C in 24 months — reflects both genuine technical progress and sector-wide multiple expansion driven by institutional conviction in physical AI. The company released a credible series of state-of-the-art models on a roughly six-month cadence (π0, π0.5, π*0.6, π0.7), open-sourced base weights that drove measurable academic adoption, and demonstrated real commercial deployments with Weave Robotics and Ultra. These are tangible milestones, not vaporware.

That said, the valuation is almost entirely a bet on future revenue, not present cash flows. Revenue as of mid-2026 is minimal — the company is in early-stage commercial deployment rather than at scale — meaning investors are pricing in a scenario in which Physical Intelligence's foundation models become the dominant control-plane for millions of industrial and service robots over the next decade. A global robotics market projected at $280 billion by 2034 provides the long-range addressable market backing the multiple.

The risk is real. Skild AI closed a $1.4 billion round at a $14 billion valuation in January 2026 — a higher valuation than Physical Intelligence's most recent confirmed round — demonstrating that the category is not winner-take-all. Google DeepMind's Gemini Robotics program, with access to virtually unlimited compute and proprietary data from Alphabet's product suite, could commoditize the foundation layer faster than Physical Intelligence can commercialize it. The open-source release of π0 weights accelerates research but also makes the model architecture easier to replicate and extend by competitors.

Physical Intelligence is not profitable and is not expected to be profitable in the near term. The company's primary cost center is compute infrastructure for training large VLA models — a capital-intensive activity with costs analogous to frontier LLM labs. With $1 billion-plus raised and minimal current revenue, the company is in deliberate investment mode: prioritizing model capability and commercial partnership expansion ahead of near-term monetization.

No IPO is expected in the next 12–24 months. The company has actively avoided giving commercialization timelines and is focused on establishing product-market fit and scale deployments before accessing public markets. The Series C round (if closed at reported terms) provides a further $1 billion runway, eliminating near-term liquidity pressure. The presence of T. Rowe Price as a crossover investor on the cap table from Series B is a structural indicator that the company is beginning IPO readiness posture — such investors typically invest 18–36 months before a public offering.

A plausible IPO window opens in 2028–2030 if fleet deployments scale to meaningful recurring revenue. Acquisition by a strategic buyer (Alphabet, Amazon, NVIDIA, or a major industrial automation conglomerate) cannot be ruled out; CapitalG's lead on Series B creates at least a conversational relationship with Alphabet, the most natural strategic acquirer given the DeepMind research lineage.

For vendors selling into Physical Intelligence, over $1 billion in raised capital means the company has genuine budget authority despite having minimal revenue. The primary spend categories are clear from the company's technical profile: cloud GPU compute for training multi-billion-parameter VLA models (H100/H200 clusters at scale), hardware procurement for the robotic arms and test platforms used in in-house data collection, and elite engineering talent acquisition — the company grew from ~80 to ~200 employees in roughly 12 months.

Priority spend areas where vendors can win: GPU cloud infrastructure (H100/H200 clusters on GCP, AWS, or Azure — the JAX framework signals significant GCP and TPU usage); MLOps and experiment tracking tooling; data annotation and teleoperation infrastructure for dexterous manipulation; and enterprise security and compliance tooling as commercial deployments scale. The company's open-source trajectory (ALOHA, DROID fine-tuning, FAST tokenizer) also creates opportunities for tooling vendors that can support research-grade workflows at scale.

As a research-culture company with a founder-led procurement process, Physical Intelligence evaluates tooling rigorously and is comfortable with open-source alternatives — pure SaaS without a research-grade option will struggle to land. The fastest commercial path in is through a technical champion on the research or engineering team, or through a warm introduction from a shared investor (Thrive, Sequoia, Lux, CapitalG). Watch for VP-level hires in sales and GTM as the signal that the procurement door is opening more formally.