When AI Grows Hands and Feet, the World Needs Taiwan Again
The rise of Physical AI is sending venture capital back to the world of hardware, manufacturing and the global supply chain, and in response TechCrunch Disrupt has put Taiwan on the Breakout Stage.
There is a shift afoot as AI leaves the screen for factories, vehicles, drones and other smart devices; capital follows suit. The coming contest in AI will be decided less by who wields the superior model and more by the ability to produce intelligent products at a global level.
As one of the preeminent early stage startup gatherings in the U.S., TechCrunch Disrupt is devoting part of its agenda this year to the subject of Physical AI. Taiwan’s presence is warranted given the surge in venture funding for R&D and hardware in the last two years. Having long been the force behind global AI chip production, the island is once more of strategic import. Its command of semiconductors, top tier supply chains and advanced manufacturing are what underpin the work needed to make robotics and other manifestations of Physical AI a reality.
For this reason, TechCrunch has put out an invitation to Taiwan for the Breakout Stage. ITRI and Phison are set to put on a session titled “Taiwan Powering the Next AI Revolution: Connect, Invest & Scale” It is an opportunity for technology leaders, founders and investors to see what makes Taiwan such a vital platform for the coming era of AI.
We would not call this just another talk on artificial intelligence. At TechCrunch we have been focusing more on Real World AI, the kind that goes past software to run the machinery in a factory or home, autonomous systems, vehicles and infrastructure. One could say that shift is altering the very economics of how a tech company is put together.
Capital is rekindling its affection for hardware. But make no mistake, software is not disappearing.
For the better part of a decade, Silicon Valley had a preference for ventures that could put down roots with minimal physical infrastructure. The formula was simple: write some code, host it in the cloud and get customers on board. A startup of any ambition did not need to be encumbered by a factory or a complex logistics operation to grow at pace.
Physical AI has altered the calculus.
Algorithms will only take you so far once AI is tasked with piloting vehicles, operating robotic arms, flying drones or making calls in the real world. To do that, a company requires more than software. It needs the whole stack: sensors, controllers, power systems, memory and processors, to say nothing of the mechanical engineering, thermal management and supply chains necessary to go from hundreds of prototypes to hundreds of thousands of finished goods.
Jensen Huang, CEO of NVIDIA, put it plainly when he said the next frontier for AI is physical. He has even likened the coming of age for general purpose robotics to a ChatGPT moment.
Yet there is an imbalance. “Today’s robotics systems are incredible brains trapped in weak, incapable bodies,” says Haomiao Huang, cofounder of Matter. The intelligence of the machine is outstripping the machine itself. That is where the new investment is being made.
The Money Is Already Moving
You can see this in the numbers; it is more than a change in the talking points.
According to Crunchbase, global robotics startups pulled in some $15 billion in 2025, a record figure. By June of 2026, funding for the sector had climbed to around $18.8 billion, besting the entire previous year and the $14.1 billion on offer during the 2021 venture boom.
Individual rounds tell their own story. Saronic came away with $1.75 billion. Neura Robotics and Skild AI each secured $1.4 billion in financing. Apptronik’s Series A swelled to over $935 million while Mind Robotics has drawn in hundreds of millions in quick order.
This is not a matter of a few investors making a speculative wager on robots. It is a repricing of the notion that AI must eventually leave the confines of the computer screen. CB Insights puts global venture funding at nearly $469 billion for 2025, the most since 2022, with AI responsible for close to half of all dollars spent.
AI Is Becoming an Infrastructure Business
There is an even bigger shift at work upstream. Ten years ago, a tech startup’s worth was in its people and its code. Now it is in compute, GPUs, data centers and the like.
Look at Lambda, which raised in excess of $1.5 billion in a 2025 Series E to put up gigawatt scale AI infrastructure. Or Crusoe, whose latest round put a valuation on the company of over $10 billion. Then there is CoreWeave. In 2025 alone the company put more than $10 billion in cash toward property and equipment for servers and networking, with performance obligations topping $60 billion by year end.
So the questions for investors have evolved. It is not just about who has the superior model. The focus is on who can build and scale the physical infrastructure for AI, who can manufacture it and drive down the cost, and who can make a prototype a global product.
That is what makes Taiwan so critical.
What the Green Line Really Means
Our VCs WANT ROBOTS graphic uses 2006 as a baseline to chart the broad movement in venture interest over 20 years. You will see the curve move from the software era and then pick up speed with the advent of Physical AI to an estimated 480 for 2026.
Do not read too much into the 480 figure. It is not $48 billion or $480 billion in reported investment. Sparknify created the index to show market direction. And the data bears it out: robotics funding hit $15 billion in 2025 and was already at $18.8 billion by mid-2026.
Venture capital is not done with software, but it is widening the scope of what an AI company should own. The subtext of VCs want robots is straightforward: they have a need for hardware again.
Why Taiwan Matters Now
The things Physical AI demands are things Taiwan has been putting in place for decades. The U.S. Department of Commerce will tell you the island is home to over 60 per cent of global semiconductor foundry revenue and the vast majority of the world’s most advanced chips, an industry that brought in more than $165 billion in 2024.
But the semiconductors are just the start.
Taiwan’s depth in PCBs, precision machinery, ODM and OEM manufacturing, testing and mass production is hard to match. These may not have the cachet of a foundation model or a humanoid robot, but they are what separate a commercial success from a mere demo. The shift in Silicon Valley from Generative to Real World AI is changing the dynamic for Taiwan. It is no longer just the manufacturing arm of the tech industry; it has the potential to be a scale platform for Physical AI.
ITRI: Taiwan Did Not Suddenly Appear in the AI Revolution
Our first point of order at the TechCrunch Breakout Session will be a straightforward one: Why Taiwan? ITRI is in a unique position to provide the answer.
One need only look at the history of the Industrial Technology Research Institute to see its foundational importance to the island’s technology sector. In 1987, ITRI put together the semiconductor facilities, equipment and 98 specialists that would form TSMC. Before that, it was instrumental in building UMC and the wider semiconductor ecosystem.
This pedigree is significant because Taiwan’s standing in AI today is not an overnight phenomenon. It is what you get after decades of putting money into research, engineering, commercialization and the supply chain.
But at TechCrunch Disrupt, ITRI will be less concerned with past accomplishments than with what investors ought to be looking for in the future. The institute will put a fine point on the upswing in Physical AI investment and make the case for how international venture firms and corporate players can tap into Taiwan’s capabilities to speed up global expansion.
Where Are the Opportunities?
A question of more immediate interest to the investor is where the capital should be put. We will put aside the macroeconomics to spotlight a new crop of Taiwanese startups in the Physical AI space.
Expect to hear about Morale AI with its agents for smart manufacturing, or JmemTek in post quantum security. These are part of a larger thesis.
Physical AI is more than humanoid robots. It is the whole stack needed to put artificial intelligence in the physical world, from simulation and sensing to compute and infrastructure. The big opportunity for an investor may not be in guessing which robot will come out on top, but in backing the companies that form the underpinnings of the entire economy.
Phison: The Next AI Revolution May Move From the Data Center to the Edge
What comes next? Phison will tackle that from an infrastructure standpoint.
You cannot send every bit of data back to the cloud when AI is in your vehicles, factories and personal devices. Issues of latency, bandwidth, cost and privacy will force intelligence to the edge. While large data centers are here to stay, they will not be the sole home of AI.
Phison is already moving beyond its traditional work in NAND controllers and storage to address this. With aiDAPTIV, the company is using flash memory to give AI systems the effective memory they need to run larger models locally. It is a kind of decentralization not unlike what happened with the personal computer in the 1980s. When intelligence is in every machine and device, the hardware and storage matter. That is something Taiwan knows well.
Bringing Taiwan Onto the Global Physical AI Stage
Sparknify has put together this session for a reason. We do not want to put on another event to explain Taiwan; we want to put Taiwan at the heart of one of Silicon Valley’s key investment dialogues.
As global VC looks for its next move in robotics and AI infrastructure, Taiwan should be seen as more than a place of manufacture. It can be a platform for scaling. Silicon Valley has the software, the talent and the capital. Taiwan has the semiconductors and the most sophisticated supply chains. If the task of the coming decade is to turn AI models into millions of real world products, the link between the two will be vital.
We are coordinating with ITRI, Phison and the TechCrunch platform to deliver that message to the right people.
See You at TechCrunch Disrupt on October 13
For those of you making plans for TechCrunch Disrupt this year, add this to your agenda and be there early. Capacity is limited for our Breakout Sessions.
“Taiwan Powering the Next AI Revolution: Connect, Invest & Scale” is on the Breakout Stage at Moscone West in San Francisco on October 13, 2026, from 1:30 to 2:20 PM as part of the Robotics and Physical AI program. ITRI and Phison will be on hand to discuss why Taiwan matters and where the opportunities lie.
Should you still need a pass for the main venue, head to the TechCrunch Disrupt page below and use promo code JOINME for a 25 percent discount on your ticket at
Generative AI set out to teach machines to think. Physical AI will put them in the world. And when AI calls for chips, factories and supply chains, the eyes of the world will once again be on Taiwan.
Comments