When the foundation changes, the giants fall

The semiconductor industry consumes as much water as Hong Kong. The technology that replaces most of that process already exists.

I studied physics at the University of Oulu in the late 1980s. One of my classmates was Johannes Frantti. We went in different directions. I spent decades building and rebuilding companies in microelectronics and precision dispensing. He spent decades in materials science. Recently, I reconnected with his work, and it stopped me in my tracks.

Johannes has developed a thin-film material that collapses chip patterning from several hundred process steps to two. Grow the film. Write the pattern directly into it. No photoresist. No mask set. No long chain of deposition, lithography, and etch cycles. The material is a bulk material. Low cost. Stable. Free of rare earth elements.

To understand why that matters, you need to understand what happens inside a semiconductor fab today.

You grow an ultrapure silicon crystal, slice it into wafers, polish them perfectly flat. Deposit thin insulating and conducting films onto the wafer. Coat the surface with photoresist. Project the circuit pattern through precision masks using extreme ultraviolet light. Etch away the exposed material with aggressive chemicals or plasma. Implant dopant ions to tune electrical behaviour. Repeat those steps dozens of times to stack up the layers of a finished chip. Add metal interconnects, planarize repeatedly, then test, dice, and package.

What sounds like eight steps is, in reality, several hundred individual process steps. Each one requires its own tools, masks, gases, and chemicals. It is breathtaking engineering. It is also breathtakingly resource-hungry.

The industry already uses roughly 8 litres of water and over 1 kilowatt hour of energy per square centimetre of chip. Approximately 0.84 kilograms of CO2-equivalent emissions per square centimetre. Perfluorocarbons have atmospheric lifetimes measured in thousands of years. PFAS, the so-called chemicals forever, are woven through the entire process. Wastewater carries heavy metals and volatile organic compounds.

Then there is the materials problem. A modern chip draws on gallium, germanium, cerium, europium, gadolinium, lanthanum, neodymium, praseodymium, scandium, terbium, yttrium, and others. China accounts for roughly 98 per cent of global raw gallium production and about two-thirds of raw germanium. In 2023, it imposed export controls on both. Not a theoretical supply chain risk. A demonstrated one.

Frantti's approach sidesteps most of that chemistry, most of that water burden, and the rare earth dependency in a single move. The capital expenditure savings are enormous. It opens chip-scale fabrication to companies that could never afford a traditional fab.

But the part that changed how I think about the future is not the chip itself. It is diagnostics.

Most smart assays today are barely smart at all. They carry a few conductive patterns to read current or changes in capacitance. Functional, but primitive. The intelligence stops at the edge of the chip because putting real sensing onto a low-cost disposable was never economical. A material you can pattern directly, cheaply, and at scale removes that ceiling. The patterned features are embedded and conductive, defined at the micron or sub-micron scale. They can function as electrodes, magnets, heaters, sensors, or interconnects.

The market is not small. Point-of-care diagnostics is already near $ 50 billion and is heading toward $ 105 billion by 2033. Biosensors sit around 32 billion in 2025, growing at about 9 per cent a year. Microfluidics is at 24 billion and growing at double-digit rates, with lab-on-a-chip the largest segment.

A low-cost, directly patterned, intelligent assay lands almost everywhere the test strip and the lab tube live today. Infectious disease, still roughly 30 per cent of the point of care revenue. Glucose monitoring, close to 38 per cent of the market. Cancer and biomarker testing, where sensitivity and cost per test decide everything. Wearable and continuous monitoring, a category that lives or dies on cheap integrated sensing. Environmental monitoring, food safety, and agriculture. Every one of them is constrained by the same requirement: the assay must be smart enough to be useful and cheap enough to be disposable.

Imagine diagnostic devices as intelligent as a microchip, priced like the plastic cartridge they replace. Single-use, smart, affordable diagnostics. Finally made possible because the manufacturing allows it.

Kodak had every advantage when digital imaging arrived. It did not matter. The foundation underneath had changed, and the old way of doing things stopped making sense almost overnight.

I intend to help build what comes next.

Kauko Väinämö is CEO of Ginolis