Home / Company / Blog / The Trillion-Touch Opportunity: Why Machines That Can Feel Are the Next Great Market

The Trillion-Touch Opportunity: Why Machines That Can Feel Are the Next Great Market

Part 8 and the final installment of The Missing Sense, a series on Physical AI, tactile sensing, and the road to dexterous machines. 

 

Over the past seven posts, our technology experts have taken you inside a single robotic fingertip: the physics of a slip, the craft of an elastomer, the intelligence that turns contact into meaning, and the unglamorous problems that decide whether any of it survives the real world. For this final installment, I want to pull back from that single fingertip and ask the question we’ve been building toward: what changes when robots can truly feel what they touch?

The remarkable thing about touch is what happens when you multiply it. One robotic fingertip is an engineering problem, but a billion of them is an industry problem. As Physical AI moves machines from seeing and hearing the world to handling it, touch is set to become one of its most essential, and most multiplied, components. 

The Wave That’s Coming Needs Hands 

Automation is entering a new phase. Industrial robots, cobots, and warehouse grippers already work alongside people, and a new generation of humanoids and specialized machines is joining them. As Physical AI moves these machines from repeating fixed motions to adapting to the world around them, the number that physically handle things is set to grow dramatically.  

Every single one of these machines runs into the same wall we opened this series with: the bottleneck isn’t the brain or the legs; it’s the hands. A robot that can plan a task flawlessly but still drops the cup or crushes the grape isn’t a product. It needs dexterity, and you can’t have real dexterity without feeling contact, force, and slip. That’s the missing sense.

That makes tactile sensing non-optional for any machine that handles the physical world. Whatever design wins, it will need to feel, whether it’s a five-fingered humanoid hand or a simple gripper on a wheeled base. That’s why we’ve built Physical AI solutions that pair our touch controllers with Astra® Edge AI processors. 

Count Contact Points, Not Machines 

The unit of demand isn’t the machine. It’s every point where that machine needs to sense and respond to physical contact. That’s what makes this market larger than it first appears.  

A dexterous hand may carry dense sensor arrays on every fingertip, along the fingers, and across the palm. A humanoid adds a second hand, feet that need to feel the ground for balance, and eventually skin for safe contact with people working nearby. As an example, our N6012T capacitive touch controller IC packs 60 sensing channels into a single array. Put one on each of ten fingertips and a single humanoid carries 600, before you count palms, feet or skin. Add the replacement skins that wear out and the data-collection gloves used to train the next generation, and the necessary components needed for tactile sensing runs hundreds of times larger than the machine count.

There’s a nice symmetry buried in that. In Part 4, Derek Solven described how Synaptics has shipped more than two billion touch controllers, which together have registered well past a trillion individual touch events, one fingertip on glass at a time. A working robot hand touches the world constantly, gripping, adjusting, and catching slips shift after shift, and every machine put to work further increases the count. The trillion touches behind us turn out to be the training for the trillions ahead.

At Scale, Economics Decide the Technology 

A lab demo can afford a sensor that is expensive, bulky, and power-hungry. A high-volume robotics program can’t. Across every contact point on millions of machines, each extra dollar, milliwatt, and wire multiplies until it dominates the bill of materials. So the real question is which approach delivers trustworthy tactile intelligence per dollar, per cubic millimeter, per millisecond. As Javier Del Prado laid out in Part 7, the cleanest road to the mass market belongs to compact, low-power approaches whose weaknesses are tractable, such as a thin capacitive array paired with a fast dynamic channel. The market, more than any lab, will settle it. 

Beyond Robots: Prosthetics and Smart Skin

The opportunity reaches beyond the factory, and the example that moves me most is prosthetics. At the Synaptics Tech Day in October this year, Dr. Aadeel Akhtar, founder and CEO of PSYONIC, reminded us what’s at stake: PSYONIC’s touch-sensing bionic technology is improving the quality of life for the people who use it, and it’s now helping robots adapt to the world around them. Further out, large-area skin could let robots work safely alongside people.

What It Takes to Win in Robotic Touch  

Plenty of companies have a promising piece of the puzzle, whether a skin, sensor, or algorithm. The hard part is bringing the whole stack together and making it survive the real world: sensing ICs, the fusion of fast and slow channels into one coherent sense, Edge AI that turns raw signal into clean events right at the hand, materials that stay reliable for millions of cycles, and the manufacturing and partnerships to deliver it at volume. Underneath it all is understanding how real-world sensing fails, which you only earn by shipping billions of units.

That’s a rare combination, and it takes decades to build. At Synaptics, we’ve spent forty years building silicon that senses: first on touchpads, then on phones, laptops, and cars, and now on the fingertips of machines. Our founders, Federico Faggin and Carver Mead, named the company in 1986 after the junctions between neurons, betting that silicon could one day sense the way living things do. Forty years later, the robotics industry is asking for precisely that. 

The Missing Sense, Found 

Satish Ganesan opened this series with a simple act: reaching into your pocket and finding your keys without looking. Vision gave robots a view of the world; touch is what will finally let them reach into it. Whoever delivers reliable touch at scale will supply a foundational layer of the age of dexterous machines.

If a robot can anticipate a slip and adjust its grip in real time, what becomes possible?

Let’s find out together. If you’re building machines that need to feel, we’d like to hear from you. To go deeper, explore our approach to Physical AI, or read our whitepaper, Enhancing Robotic Hands Dexterity Using Touch ICs. 

Rahul Patel

With more than 30 years of leadership experience in the semiconductor industry, Patel has a proven track record of driving growth and product innovation, particularly in the areas of high-performance Edge-AI wireless connectivity solutions for handsets, tablets, PCs, wearables such as smartwatches and earbuds, IoT applications, and networking and broadband solutions for enterprises and home markets.

Read more by Rahul Patel
AI Edge Computing
Receive the latest news