TSMC’s 1.6nm Process Has Received Orders From Apple and OpenAI

TSMC’s new generation chip A16, which uses angstrom-level process technology, is still nearly two years away from its expected mass production time. However, the chip has won the favor of many industry giants in advance. According to a report by Taiwan’s Economic Daily on September 2, Apple, a major customer, has not only booked the first batch of TSMC’s A16 chip production capacity, but also, due to the long-term demand for self-developed AI chips, OpenAI has also joined the ranks of booking A16 production capacity.

As the developer of ChatGPT, OpenAI is actively engaged in the layout of self-developed AI chips to meet its growing computing power needs. Through cooperation with TSMC, OpenAI can not only obtain the most advanced process technology support, but also rely on TSMC’s excellent manufacturing capabilities to accelerate the research and development and mass production of its AI chips.

It is revealed that OpenAI originally planned to work with TSMC to build a dedicated wafer fab, but after careful consideration, the company decided to work with industry partners such as Broadcom and Marvell to jointly develop AI chips. The plan is expected to be put into production through TSMC’s 3nm series and subsequent A16 process technology. This decision not only demonstrates OpenAI’s flexibility and foresight in strategic planning, but also further strengthens TSMC’s leadership in the global semiconductor manufacturing field.

The reason why TSMC’s A16 process technology has received such widespread attention and popularity is due to its unique technical advantages and broad application prospects. As the most cutting-edge process node currently disclosed by TSMC, the A16 chip marks that TSMC has officially entered a new stage of the angstrom process and is scheduled to be put into mass production in the second half of 2026. It is worth noting that 1 angstrom is only one-tenth of 1 nanometer. In the context of the current semiconductor industry having successfully broken through the 2-nanometer process technology, the angstrom process has undoubtedly become a new high ground for the world’s top chip companies to compete for.

According to the information released by TSMC, the A16 process will adopt the innovative next-generation nanosheet transistor technology and integrate TSMC’s Super Power Rail (SPR) architecture. SPR technology cleverly transfers the power supply line to the back of the wafer, thereby freeing up more space on the front of the wafer for the layout of signal lines, thereby improving logic density and performance. In addition, SPR can effectively and significantly reduce IR drop (voltage drop), thereby improving the overall power supply efficiency.

TSMC particularly emphasized that the backside contact technology used in the A16 process can maintain the same gate density, layout footprint and component width adjustment flexibility as the traditional front power supply, and show its unique advantages in high-performance computing (HPC) products. It is worth noting that the backside powered network technology (BSPDN) was originally planned to debut in TSMC’s N2P node, but with the adjustment of production planning, the technology has been removed from the N2P node and has become a highlight of the A16 process.

Compared to TSMC’s N2P manufacturing process, the A16 is expected to deliver an 8% to 10% performance improvement at the same voltage and complexity, or a 15% to 20% power reduction at the same frequency and number of transistors. Although TSMC has not yet announced specific density parameter details, the company revealed that the density of the A16 chip will increase by 1.07 to 1.10 times. This series of significant technical indicator improvements undoubtedly indicates that TSMC’s customers will be able to enjoy more excellent product performance and energy efficiency.

For OpenAI, this chip using the A16 process will greatly enhance Sora’s video generation capabilities and is expected to be integrated into Apple’s generative AI feature suite, thereby helping Apple’s product sales to further increase. This cooperation with TSMC marks that OpenAI is no longer limited to software development, but has begun to actively invest in the hardware field to better support the development of its artificial intelligence. It is worth noting that OpenAI did not choose to build its own manufacturing facilities, but directly cooperated with leading semiconductor companies. This decision fully demonstrates its philosophy of open cooperation. Although mass production will take some time and is expected to start as early as 2026, the market’s expectations for this chip and the technology behind it are self-evident.

In summary, the fact that TSMC’s 1.6nm process has won orders from Apple and OpenAl has far-reaching significance and impact. It not only reflects TSMC’s leading position and technological strength in the field of semiconductor manufacturing, but also foreshadows further improvements in chip performance, power consumption and density in the future. In addition, this cooperation will drive the vigorous development of AI and smart hardware, and further consolidate TSMC’s position in the global market. At the same time, it will also inspire a wider wave of technological innovation and promote the overall upgrading of the semiconductor industry.