Samsung Electronics’ foundry division has begun developing Neuralink’s fourth-generation brain implant chip since late last year based on a 4-nanometer process.
Test chips are expected to be shipped in the first half of next year, with mass production possible as early as the end of next year. Foundry refers to a business that manufactures semiconductors on contract without designing them directly. This is the first time Samsung has secured a Neuralink order.
◆ From brain to device, and now from device to brain
Neuralink is a neurotechnology company founded by Musk in 2016. Its goal is to implant chips in the human skull so people can control digital devices through thought without moving their hands or feet. It unveiled its first chip, “N1,” in 2019 and has since released up to its third-generation product in 2023. The company is currently valued at 12 trillion won.
The fourth-generation chip is different from previous versions. Until now, the chips have worked in one direction, reading signals from the brain and sending commands to devices. The new chip will also transmit information from devices back to the brain. In other words, signals will flow both ways. The company believes this approach could stimulate neural activity and open possibilities such as restoring sight to patients who have lost their vision.
◆ Why use an older 4-nanometer process for the latest chip?
One notable point is the choice of manufacturing process. Samsung’s most advanced technology is a 2-nanometer process, but Neuralink’s chip will use the more mature 4-nanometer node. The smaller the number, the more advanced the process, as it can etch circuits more densely. Yet the company has opted for a less advanced process for its most cutting-edge chip.
The reason is believed to be stability. Samsung has long worked with the 4-nanometer process, which results in fewer defects and more consistent production. A chip implanted directly in the brain allows no margin for error. The calculation is that a proven process is needed to deliver reliable volumes on time.
The supply-chain structure is also changing. Neuralink is believed to have primarily partnered with Taiwan’s TSMC through its third-generation chip. By adding Samsung for the fourth generation, it will establish a dual-sourcing system that does not rely on a single company. Some foreign media interpreted this as TSMC being pushed out, but the original report placed more emphasis on bringing Samsung in as a new partner to stabilize supply.
◆ The opening TSMC’s bottleneck gave Samsung
The background to Samsung’s opportunity is TSMC’s situation. With the AI boom driving huge orders from companies such as Nvidia, TSMC’s production lines have reached their limits. Although TSMC has responded by expanding monthly output for advanced processes, it has been unable to absorb all the incoming demand. The result is a shift in which customers seeking untapped capacity and supply diversification are turning to Samsung.
For Samsung, this is a foundation for a rebound. The company signed a 16.5-billion-dollar contract, worth about 25 trillion won, to supply self-driving chips to Tesla last year, and has also been winning AI chip orders from Nvidia and Google. The gap remains wide, however. In the first quarter of this year, TSMC held 73% of the global foundry market, while Samsung accounted for just 7%. Samsung Foundry head Han Jin-man recently said at an internal briefing that the probability of achieving profitability in 2028 is high.
As Samsung extends its reach into even the distant future business of brain implant chips, its cooperation with Musk’s camp is expanding from automobiles into the human body. Whether the foundry business can finally end years of losses will ultimately depend on yield and reliability at the mass-production stage. It remains to be seen what results Samsung will achieve with a strategy of making the most advanced chips using the most proven process.