Key Points
- Onsemi agreed to buy Synaptics for about $7 billion in all-stock — its largest deal ever.
- Synaptics holders receive 1.350 onsemi shares each, roughly a 19% premium.
- The deal targets "physical AI" — intelligence that runs locally in edge hardware.
- Onsemi expects it accretive within 18 months, with $200 million in annual cost synergies.
ON Semiconductor agreed to acquire Synaptics for about $7 billion in an all-stock deal, the chipmaker's largest acquisition ever, as it pushes beyond the data center into "physical AI" at the edge. Synaptics shareholders will receive 1.350 onsemi shares for each share they hold — roughly a 19% premium to the ten-day volume-weighted average price before the announcement. The deal is expected to close in the middle of 2027.
The logic is a full stack. Onsemi brings power and sensing chips; Synaptics adds edge-AI processors, connectivity, and human-machine-interface technology — the pieces that let a device run a model on-board rather than shipping data to a remote server. CEO Hassane El-Khoury framed the deal as assembling the full stack — power, sense, connected compute, and control — for physical AI. Onsemi expects the deal to add to adjusted earnings within 18 months, generate $200 million in annual cost synergies, and expand its addressable market by $30 billion to $243 billion by 2030.
The read: the chip M&A cycle is migrating from the GPU to the edge. Most of 2026's AI money has chased the data center — the GPUs and power plants behind the frontier models — but the next fight is over the unglamorous silicon that makes inference work inside a car, an appliance, or a sensor, where there is no cloud to fall back on. Onsemi writing its biggest-ever check for the "power, sense, compute and control" layer is a bet that intelligence moves out of the server rack and into the physical world, and that owning that layer end-to-end beats selling one component of it.
It is the same edge thesis Qualcomm is chasing in post-smartphone devices, and a companion to the specialization Google set off by splitting its TPU into training and inference chips. For readers mapping where durable value sits, it is another reason the case for owning the AI infrastructure stack runs all the way down to the sensor.
Source: CNBC.
Comments