The Market Is Bigger Than The Part

Every market forecast is an argument about a single input, dressed up as a table. Electronic skin by 2028 is put at USD 6.7 billion, against USD 3.2 billion in 2025, a compound rate of 24 per cent. That headline hides four segments compounding at four different speeds. Wearable health moves from USD 1.5 billion to USD 2.5 billion at 18 per cent. Robotics tactile sensing moves from USD 500 million to USD 2 billion at 40 per cent. Consumer haptics moves from USD 800 million to USD 1.5 billion at 22 per cent. Industrial and other applications move from USD 400 million to USD 700 million at 20 per cent. Read the total and you learn nothing. Read the spread and you learn where the compounding is. The market is bigger than the part. That is the first thing a component company gets wrong.

The biggest segment in 2028 is not robotics. It is wearable health monitoring at USD 2.5 billion, and it was already the biggest in 2025 at USD 1.5 billion. Inside it the pieces are uneven. Cardiovascular monitoring goes from USD 500 million to USD 1.2 billion on continuous blood pressure and arrhythmia detection. Sports and fitness wearables go from USD 800 million to USD 1 billion. Elderly care, mostly fall detection, goes from USD 200 million to USD 500 million as the population ages. Post-surgical monitoring goes from USD 100 million to USD 300 million, bought by hospitals trying to shorten inpatient stays. Neonatal monitoring goes from USD 100 million to USD 200 million, on nothing more exotic than a sensor that does not need a wire attached to a newborn. The pattern in every row is the same: the sensor is a component, the monitoring service is the product. Five sub-segments, five buyers, one supplier of the sensing layer underneath them all. Whoever prices at the component and stops there is describing a business they are not in.

Robotics tactile sensing is the smallest of the four segments at the start and the fastest at 40 per cent. USD 500 million becomes USD 2 billion in three years. The pieces again. Humanoid robots go from roughly USD 50 million to USD 800 million. Industrial robot grippers go from USD 200 million to USD 500 million on assembly automation. Collaborative robots go from USD 100 million to USD 400 million, driven less by capability than by safety requirements around people. Prosthetic sensors go from USD 50 million to USD 150 million on advanced bionic hands. Surgical robots go from USD 100 million to USD 300 million, on tactile feedback for minimally invasive work where the surgeon cannot feel what the instrument feels. A segment can be small in absolute revenue and still be the only one that decides who owns the category. Rate of change and size are two different questions, and a spreadsheet sorts by the wrong one by default. The smallest line on the table is the one that moves.

Everything above about robotics depends on humanoid unit volume, and nobody knows that number. The published scenarios are wide. Five thousand humanoid units shipped by 2028 and the robot tactile market lands near USD 500 million. Twenty thousand units and it lands near USD 1.2 billion. One hundred thousand units and it lands near USD 2.5 billion. Same technology. Same sensor. A five-fold spread in outcome, decided by a manufacturing ramp that has not happened yet. That is the honest shape of this market. It is not a forecast, it is a bet on one number, and the number is not about sensing at all. It is about whether anybody builds the robots. A supplier who plans against the middle scenario is running the middle scenario of somebody else's factory schedule. The cost of that assumption lands on one balance sheet, and it is not the robot maker's. The probability that humanoid volume arrives on the fast scenario is not zero. Most planners are pricing it at zero, which is the same error with the sign flipped.

Read the total and you learn nothing. Read the spread and you learn where the compounding is.

The cost curve is the last input, and it is the one a supplier actually controls. A basic hand with ten contact points costs USD 50 to USD 100 when tethered, USD 10 to USD 20 in production at 10,000 units a year, and USD 3 to USD 5 at 100,000 a year. A functional hand at 100 taxels runs USD 200 to USD 500 tethered, USD 50 to USD 100 at 10,000 units, USD 15 to USD 30 at 100,000. A high-resolution hand above 1,000 taxels runs USD 2,000 to USD 5,000 tethered, USD 200 to USD 500 at 10,000 units, USD 50 to USD 100 at 100,000. The threshold is not a technology threshold. At USD 50 to USD 100 per hand, touch becomes a standard feature. Above USD 500 it stays a premium option, which means it ships on machines counted in the thousands rather than the millions. Volume, not resolution, decides which of those markets you are actually in. A resolution race is a way of losing a cost race politely.

Here is the trap. A component company reads USD 6.7 billion, sees its sensor in every row of the table, and prices against the part. The part is the smallest number in the chain. The monitoring service is USD 2.5 billion. The robot fleet is USD 2 billion, and the grippers, cobots, prosthetics and surgical arms inside it are separate buyers with separate budgets. Consumer haptics is USD 1.5 billion. Industrial and other is USD 700 million. Set a component price against a component market and the arithmetic is respectable in a spreadsheet and wrong in the room where the decision gets made. The value sits one layer up, with whoever operates the monitoring, owns the fleet, or sells the outcome rather than the sensing. I have built companies across twelve countries and deployed 210 energy systems. I restructured a industrial group and signed a personal guarantee. The pattern I have watched repeat is a good technology priced against its own bill of materials while the margin sits with the operator two steps downstream. The probability that a supplier can hold a component price while the layer above it compounds at 24 per cent is not zero. Most are pricing it at zero. The market is bigger than the part. Price the layer where the money is actually collected.