Physical AI Raised More in Six Months Than in the Previous Three Years Combined

How can I invest in robotics companies?

You can invest in robotics through listed industrial automation stocks and robotics ETFs, or, if you qualify as a wholesale investor, through private positions in the humanoid and physical AI companies drawing most of the sector's new capital. Investors put US$47.4 billion into physical AI in the first half of 2026, more than across the whole of 2022 to 2024 combined, according to Crunchbase, which counts robotics, autonomous vehicles, drones, aerospace and industrial automation within the category. Figure AI, Physical Intelligence, 1X Technologies, Apptronik and Agility Robotics are all still private, and the listed options available to most investors reach only part of that activity.

What are the main ways I can invest in robotics companies?

There are four main routes:

1. Listed equities with robotics or automation exposure
2. Exchange-traded funds built around robotics themes
3. Private company investment through venture funds or SPVs
4. Direct secondary purchases of shares in a specific private robotics company

Listed companies such as Fanuc, ABB, Keyence, Intuitive Surgical and Teradyne give exposure to established automation businesses with long operating histories, although most of their product lines were designed before the current generation of AI-driven robotics emerged. Nvidia is the most common listed proxy for the category, since its chips and simulation software sit behind the training of most modern robotic systems.

Robotics ETFs such as Global X's BOTZ bundle listed exposure across industrial automation, autonomous systems and AI hardware. They offer diversification, but because they hold only listed shares, the private companies attracting most of today's investment sit outside them.

Private company access, through venture funds or SPV structures, is where the earliest-stage robotics exposure sits, and it is also where the entry requirements are strictest.

Do I need to be a wholesale investor to invest in private robotics companies?

Yes. In Australia, private placements in unlisted companies are restricted to wholesale investors under the Corporations Act 2001, and the standard tests are net assets of at least A$2.5 million or gross income of A$250,000 a year for each of the two most recent financial years, confirmed by a qualified accountant. NonPublic's wholesale investor guide explains how each test works and what documentation you need.

The same logic applies to the offshore structures many Australian investors use. An SPV incorporated in the Cayman Islands or Delaware that accepts Australian investors will ask for evidence of wholesale status before taking a commitment, and the SEC's accredited investor rules, the UK's high net worth definitions and equivalent regimes across Europe and Asia all reserve private market participation for investors who meet defined financial thresholds, wherever the deal is structured.

The rationale is that private company investments carry risks of a different kind from listed markets. Illiquidity, limited disclosure and the absence of daily pricing mean investors need to be able to absorb a loss without it affecting their financial stability, and NonPublic's guide to private market risks covers this in detail for Australian investors.

What robotics companies can I invest in right now?

There are dozens of well-funded private robotics companies, but very few are directly accessible to individual investors, and they divide broadly into humanoid robotics, task-specific autonomous systems and the infrastructure that supports both.

Humanoid robotics has drawn the most capital and attention. Figure AI raised at a US$2.6 billion valuation in early 2024 with backing from Microsoft, OpenAI, Nvidia and Jeff Bezos, then reached a US$39 billion valuation in its September 2025 Series C. Physical Intelligence, founded by former Google DeepMind researchers, builds foundation models for robotic control rather than hardware, and was reported in March 2026 to be raising around US$1 billion at an US$11 billion valuation. 1X Technologies, backed by the OpenAI Startup Fund, is developing its NEO humanoid for homes and struck a deal in December 2025 to send it into factories and warehouses. Apptronik, whose Apollo robot grew out of its team's humanoid work with NASA, has raised more than US$935 million across its Series A at a valuation above US$5 billion.

Further along the commercial path are companies built for specific tasks, including Agility Robotics, whose Digit robots work in GXO logistics facilities, Nuro, which now licenses its autonomous driving system to partners including Uber and Lucid, and Zipline, which runs drone delivery at commercial scale. Having moved from prototypes to paying customers, these businesses carry a different risk profile from humanoid companies still in pilot programmes.

The infrastructure layer includes Nvidia, whose Omniverse and Isaac platforms are widely used to simulate and train robotic systems, along with a cluster of data companies supplying the training sets robots learn from. Covariant, which builds AI that teaches robots to handle objects, lost its founders to Amazon in 2024 when the company hired them and licensed Covariant's models, taking one of the most interesting independent players out of reach for outside investors.

Which of these companies can be reached at any given time depends on whether existing shareholders are selling and whether pre-IPO allocations are being structured, so availability shifts from month to month. Look at the opportunities currently available on NonPublic.

What is physical AI and why should investors care?

Physical AI refers to machine learning systems designed to sense and act in the physical world, as distinct from software AI that operates entirely in digital environments, and investors should care because the two create value in very different ways.

Software AI products like ChatGPT or Gemini can be copied at almost no marginal cost once built, which is part of why software companies command premium revenue multiples. Physical AI products involve hardware that costs money to manufacture, supply chains that add operational complexity and robots that wear out and need maintenance, so their unit economics sit closer to industrial machinery than to software, although the intelligence layer creates room for performance gains and platform lock-in that traditional industrial equipment has never offered.

NonPublic's physical AI analysis traces how the category has grown from almost nothing a decade ago to a combined private company valuation above US$263 billion. That growth has tracked the improvement in AI models and the falling cost of sensors and compute, which together have made capable autonomous systems economical to build at commercial scale for the first time.

For investors, the turning point is the shift from robotics as a custom, project-based business to robotics as a product with repeatable unit economics. Agility's Digit units working in GXO warehouses and Figure's testing at BMW's Spartanburg plant represent the early stages of that transition, and the companies that establish defensible positions before commoditisation pressure arrives will determine which investments in the category pay off.

How do I value a private robotics company?

Robotics companies are valued using different frameworks depending on their stage and business model. Early-stage companies without commercial deployments are typically valued on milestones, taking in what has been demonstrated technically, how large the addressable market looks and what the team's prior work suggests about its ability to execute, since comparable company multiples offer little guidance at that point.

Companies with commercial deployments are usually valued on revenue multiples, with the peer group drawn from industrial automation businesses such as Fanuc and ABB, high-growth software companies for the AI platform component, or some blend of the two depending on how revenue is structured. A robotics company earning recurring software and maintenance revenue commands a higher multiple than one selling hardware outright, so understanding the revenue model comes before applying any multiple.

Strategic investors also shape private robotics valuations. When Nvidia, Amazon, Toyota Ventures or SoftBank invest in a robotics company, they bring capital along with a signal about the technology and potential customer relationships, and NonPublic's H1 2026 analysis shows how capital concentrated around a small number of strategically backed companies in the first half of 2026.

What are the risks of investing in private robotics companies?

The largest risk is scaling hardware. Producing and deploying physical robots at commercial volume involves supply chain management, quality control and unit cost discipline that software companies never face, and several well-funded robotics companies have produced impressive prototypes and then struggled to manufacture at a cost that makes commercial sense.

Regulation is a risk specific to this category as well. Autonomous systems working in warehouses, on roads or in healthcare settings face regulatory frameworks that vary by jurisdiction and can change quickly after an incident, and the rules for humanoid robots working alongside people are still being written in most countries.

Talent concentration is a sharper risk here than in most sectors, because the leading researchers in physical AI form a small community and the departure of even a few of them can be hard to assess from outside, which makes the stability of the founding team worth examining closely over the life of any investment.

Liquidity timelines are long. Even companies with commercial deployments are likely several years from a public listing, and for those still in early development the wait extends further, so capital committed to private robotics should be treated as illiquid for five years or more and sized accordingly.

Is there a listed alternative to investing in private robotics companies?

Yes, although the exposure is different. The most direct listed exposure to AI robotics is Nvidia, whose growth has been driven substantially by demand for AI training and inference, including the simulation workloads behind robotic development, and whose Project GR00T foundation model for humanoid robots is a direct bet on the category from the infrastructure side.

Beyond Nvidia, listed industrial automation companies such as Fanuc and Yaskawa in Japan, ABB in Switzerland and Rockwell Automation in the United States offer exposure to automation more broadly. They remain the bellwethers of the sector, but their product lines were built for programmed, deterministic automation rather than the adaptive, learning-based systems that define the current generation of physical AI.

The gap between listed and private robotics exposure has widened as the most closely watched companies have stayed private for longer and at higher valuations. That gap is the case for private market access in this category, and it is also where the wholesale investor requirement comes into play, since listed alternatives cover the broader automation sector without reaching the specific companies where the current growth is concentrated.

NonPublic Pty Ltd (ABN 49 607 216 928) holds Australian Financial Services Licence #482668. Investments are available to wholesale and sophisticated investors as defined under the Corporations Act 2001. This content is general in nature and does not constitute financial product advice. It does not take into account your objectives, financial situation, or needs. Investing in private markets involves significant risk, including the potential loss of your entire investment. Past performance is not a reliable indicator of future results. You should obtain independent financial advice before making any investment decision.