Insights from Dan McGrath, TechInsights at ISE 2026
At first glance, image sensors can feel like a mature technology. Resolution increases, incremental improvements, familiar players. From a market perspective, the story appears stable.
But as Dan McGrath of TechInsights made clear at Image Sensors Europe 2026, that stability is misleading.
“Under the hood, there’s tremendous differentiation—and constant surprises.” — Dan McGrath, TechInsights
From scaling to functionality
For decades, image sensor development followed a clear trajectory: shrink pixels, improve cost, increase resolution. That path has now slowed; not because progress has stopped, but because it has moved elsewhere.
“If you stop scaling pixels, where do you put the technology?” — Dan McGrath, TechInsights
The answer is increasingly inside the pixel itself.
Modern designs are adding functionality through more complex pixel architectures: enabled by structures such as embedded capacitors and multi-transistor designs. The classic 4T pixel is no longer the endpoint; more complex configurations are emerging to support features like temporal binning, improved dynamic range, and advanced readout schemes.
In parallel, different sensor types such as CMOS image sensors and SPAD-based devices are converging in terms of integration challenges, even as they serve different applications. Both benefit from stacking and both face similar questions about where innovation should sit in the structure.
Rethinking optics, materials, and spectrum
Innovation is also happening before light even reaches the photodiode.
Traditional stacks built around microlenses and colour filters are now being challenged by nanostructures (an area explored by companies such as Eyeo, whose approach replaces filters entirely with glass-like nanostructures on top of the sensor) alongside metasurfaces and plasmonic techniques. These techniques allow light to be steered rather than simply filtered, opening the door to more efficient colour separation and new multispectral capabilities.
At the same time, sensors are expanding beyond the visible spectrum. From near-infrared optimisation to ultraviolet-sensitive designs and exploratory materials such as quantum dots, image sensors are increasingly being engineered for specific spectral tasks, not just general-purpose imaging.
Not all innovations persist, however. As McGrath noted, some ideas, such as more complex backside structures, have appeared in development but ultimately given way to simpler, more manufacturable solutions. Progress is not linear; it is shaped as much by process constraints and cost as by technical possibility.
From devices to systems
If pixels are becoming more functional, sensors themselves are becoming more integrated.
Stacked architectures already allow image sensors to combine optimised pixel processes with advanced logic nodes. But the next shift is the gradual movement of computation closer to the sensor: into the image signal processor and, increasingly, into neural network structures embedded at the edge.
Reverse engineering data shows early examples of this transition, with neural network-enabled ISPs introducing more memory, more logic utilisation, and even encroaching into the pixel array itself.
This raises a key question for the industry:
will image sensors become fully intelligent edge devices, or remain part of a distributed system spanning sensor, processor, and cloud?
A fragmented future
Perhaps the most important takeaway is that image sensing is no longer following a single roadmap.
Rather than a continuous scaling curve, McGrath’s data shows a fragmented landscape, where different applications occupy distinct regions defined by pixel size, cost, and functionality.
Smartphones, automotive, scientific imaging, and emerging sensing applications are each evolving along their own trajectories, driven by different constraints and priorities.
Looking ahead
After decades defined by scaling, image sensors are entering a new phase: one shaped by functionality, integration, and application-specific design.
The challenge now is not just technological. It is strategic.
As McGrath emphasised, innovation must sit at the intersection of what is possible in process technology and what makes sense in the market. The companies that navigate that balance best will define the next generation of imaging.
