Physical AI investment is expanding beyond humanoid robots into technologies that help machines operate in factories, assist surgeons and interact with the physical world. For Amulya Vishwanath, a former Nvidia executive who founded the early-stage investment firm Techable VC, the opportunity lies in identifying the foundational technologies that can turn scientific advances into working products.
The firm has already invested in eight early-stage companies across robotics, materials science and AI-enabled life sciences, according to an August 2026 profile published by UC Berkeley Haas. Its portfolio includes companies developing industrial robotic skills, home-assistance robots and intelligent surgical systems.
Vishwanath’s move from Nvidia to venture capital reflects a shift in how she applies her technical experience. Rather than helping established AI platforms reach developers and customers, she now backs founders attempting to build the next generation of physical technology. Her investment strategy also raises a question for the wider industry: which parts of physical AI will create lasting businesses, beyond the robots that attract the most attention?
From Nvidia’s AI Ecosystem to Early-Stage Investing
Vishwanath spent seven years at Nvidia working across technical product marketing, developer relations and robotics. Her role involved helping developers and businesses adopt AI platforms, giving her exposure to the process through which technical capabilities become products used outside research environments. She subsequently founded Techable VC after completing the executive MBA programme at UC Berkeley Haas in 2025.
The transition is significant because evaluating an early-stage technology company requires a different perspective from marketing an established platform. Investors must assess whether a technology can work reliably, whether a team can develop it into a commercial product and whether customers will pay to use it.
Vishwanath’s engineering background adds another dimension. She holds a master’s degree in electrical engineering, with a focus on circuit design and microelectronics, according to Berkeley Haas. That technical foundation is relevant to investments in which the underlying science, hardware and software must work together before a product can reach the market.
Her approach is therefore centred on founders working through complex technical problems, rather than treating AI as a feature that can simply be added to an existing business.
Why Physical AI is More Than Robotics
The term physical AI often brings humanoid robots to mind. Yet the technology extends to systems that interpret the physical world and help people or machines act within it. This includes industrial automation, scientific discovery, medical technologies and advanced materials.
Techable VC’s investment focus reflects that wider definition. Its website identifies areas such as industrial robotics, neural interfaces, scientific AI, quantum materials and computational chemistry among the fields it is interested in backing.
Physical AI and Frontier Technology
The distinction has practical implications for investors. A robot is a visible product, but its performance depends on less visible capabilities: sensors that collect reliable information, software that interprets it, control systems that guide movement and hardware that can withstand real operating conditions.
The same principle applies outside robotics. An AI-enabled scientific platform, for example, may help researchers investigate a disease or analyse candidate materials. Its value depends on whether it improves the quality or speed of the research process, not simply on whether it uses an advanced model.
For venture capital, this broadens the field of opportunity while making technical assessment more demanding. Investors must understand not only what a product does, but also which scientific or engineering barrier it overcomes.
Five Investments Reveal How Techable VC Approaches Physical AI
Techable VC’s early portfolio offers a more concrete view of its investment thesis than the broad category of physical AI alone.
1.Trener Robotics: Making industrial robots more adaptable.
Trener Robotics is working on adaptive intelligence for industrial robots in real-world production environments. The difficulty is to make robots get on with useful work as reliably as possible when factory conditions change not simply to automate rigidly controlled routines.
Importance of the technology to investment: it could help automation of production tasks that are hard to operate with fixed programming.
2. Syncere: Rethinking robots for everyday assistance
Syncere is working on robotic lamps to assist in the home. This concept presents a new way to home robotics, integrating assistance into a familiar object rather than building a humanoid robot.
Importance of the investment: the design encourages interesting questions of can we do better for everyday tasks with specialized, less visible mechanisms than with versatile robots?
3. Alpha Z: Automating hazardous work
Alpha Z builds autonomous robots for repetitive or hazardous work in areas like construction, security, and industrial inspection. These environments are ideal for machines to take on tasks that are dangerous, repetitive, or difficult to staff.
Importance of investment: the technology can be evaluated by customers for practical task efficiency, safety, and lower exposure to the risk.
4. Andromeda Surgical: Bringing intelligence into surgical systems.
Andromeda Surgical has created a smart surgical navigation platform. It illustrates the sector of health and AI where the technical capability, the clinical validation, the safety and the right regulatory oversight all come together.
Importance of investment: the whole point of the medtech industry is that it has to demonstrate clinical benefit as well as be safely rolled out.
5. A stealth startup targeting brain diseases
(Reporting by Berkeley Haas) Techable VC backed a startup founded by University of California researchers building a platform to diagnose and treat brain diseases. The startup was still in stealth when that report was published and additional technical and commercial information has not been released.
Importance of investment: this demonstrates that the firm’s frontier technology definition carries over into the research-based healthcare realm, where research discovery and engineering must be commercialized.
Combined, the examples demonstrate that the firm’s idea is not confined to one product family. It includes factory automation, consumer help and hazardous fieldwork and healthcare, and all are united by the technological innovations that can be applied to the outside world.
What Vishwanath Looks for Before Backing A Founder
A promising technology is only one part of an investment decision. In its August 2026 profile, Berkeley Haas reported that Vishwanath evaluates founders against three criteria: demonstrated experience in the problem they are trying to solve, the ability to attract strong talent, and the judgment needed to navigate technical, regulatory and geopolitical complexity.
These criteria are particularly relevant to physical AI startups.
- First, domain knowledge matters. Founders need to understand the problem beyond its surface description. Building a robot for a factory, for instance, requires knowledge of production workflows, equipment constraints and the consequences of failure.
- Second, talent determines whether a difficult idea can become a product. Physical AI may require specialists across software, hardware, controls, manufacturing and scientific research. A founder must be able to assemble a team capable of working across those disciplines.
- Third, commercial judgment matters as much as technical ambition. A medical technology company must navigate clinical and regulatory requirements. An industrial robotics startup must demonstrate reliability and integrate with existing operations. Progress in either case can be slowed by obstacles that have little to do with the sophistication of its AI model.
This framework points to a wider lesson for technology investors: evaluating frontier companies requires an understanding of how the technology will be built, validated, adopted and supported over time.
The Commercial Test Comes After the Technical Breakthrough
Physical AI offers opportunities across industries, but it also faces a demanding path from prototype to deployment. A robot may perform impressively in a demonstration and still struggle with unfamiliar objects, changing environments or unexpected interruptions.
For businesses considering adoption, the relevant measures are operational. Does the system perform its assigned task consistently? How much supervision does it need? What does installation and maintenance cost? Can it integrate with existing equipment, and does it improve productivity enough to justify the investment?
Healthcare introduces additional requirements, including clinical evidence and regulatory compliance. Scientific AI faces a different challenge: whether its outputs lead to reproducible discoveries or more effective research.
These differences make it difficult to evaluate every physical AI company using a single measure of progress. A laboratory platform, a factory robot and a surgical navigation system have different users, risks and commercial timelines.
For Techable VC, backing companies across these categories means assessing each technology on its own terms while looking for founders capable of navigating the long path to adoption. The firm’s investment strategy places technical depth and execution at the centre of that process.
What Makes this Investment Strategy Worth Watching
Vishwanath’s move from Nvidia to founding Techable VC offers a case study in how experience inside a major technology ecosystem can inform early-stage investing. Her portfolio also illustrates why physical AI should be understood as a collection of related technologies rather than a synonym for humanoid robots.
The firm had invested in eight early-stage companies by the time of Berkeley Haas’s August 2026 report. That is an early portfolio, not evidence that its investment strategy has already produced long-term financial returns. Its significance lies in the range of problems it is choosing to address and the technical and commercial demands those problems create.
The next test will be whether these companies can move from promising technology to products that customers adopt, use reliably and are willing to pay for. Their progress will reveal more about the commercial potential of physical AI than the number of robot demonstrations or announcements the industry produces.
For investors and founders, the central question is therefore not simply which machine can do more. It is which underlying technology can solve a difficult problem consistently enough to become part of everyday industrial, scientific or clinical work.
Author’s POV:
What interests me about physical AI is the shift in how we measure technological progress. For years, much of the public conversation around AI has centred on what a model can generate, predict or answer. Physical AI introduces a harder test: can that intelligence perform reliably in the real world, where materials behave unpredictably, machines encounter unexpected conditions and mistakes can carry serious consequences?
Amulya Vishwanath’s move from Nvidia to venture capital offers an interesting perspective on this transition. Her investment focus spans robotics, materials science and AI-enabled life sciences, reflecting the possibility that some of the most important advances will emerge from technologies that receive less public attention than humanoid robots.
I believe the next phase of physical AI will depend as much on execution as invention. A promising prototype is only the beginning; reliability, safety, affordability and practical adoption will determine whether it becomes commercially meaningful. The real measure of progress will be the problems these technologies solve and the difference they make beyond a laboratory demonstration.
