The semiconductor industry is buzzing with anticipation as TSMC prepares to ramp up production of its cutting-edge A16 angstrom process. This advanced technology has already caught the eye of tech giant Apple, which has reportedly placed early orders. However, Apple isn’t the only one interested in leveraging this breakthrough. OpenAI is also gearing up to create a custom chip using TSMC’s A16 angstrom process, aiming to significantly enhance Sora’s video-generation capabilities.
Recent reports indicate that OpenAI was initially in advanced discussions with TSMC to develop a dedicated manufacturing facility for this custom chip. However, those ambitious plans were ultimately shelved after a thorough evaluation of profitability and other key factors.
In today’s competitive tech landscape, major companies are increasingly developing bespoke solutions to maintain their edge. An example is ByteDance, the parent company of TikTok, which has reportedly teamed up with U.S. firm Broadcom to mass-produce an in-house AI chip using TSMC’s 5nm process. Not to be outdone, OpenAI is reportedly engaged with TSMC to potentially set up a new fabrication plant specifically for its chip. Despite these discussions, the plan was eventually abandoned after weighing the economic viability.
One of the key motivations for OpenAI’s interest in the A16 angstrom process is its potential to dramatically enhance video-generation capabilities—a feature likely to attract partnerships with other industry leaders, possibly even boosting sales of Apple devices if integrated into its suite of generative AI tools.
There’s speculation that OpenAI may be collaborating closely with Apple on this venture, given Apple’s extensive expertise in developing sophisticated chipsets. However, nothing has been officially confirmed at this stage. It’s worth noting that mass production of the A16 angstrom process by TSMC is not expected to start until 2026, so the wait might be longer than some enthusiasts hope.
The anticipation seems justified; early analyses suggest that the A16 angstrom process will offer notable advantages over the next-generation N2P process. It is expected to deliver an 8-10 percent speed boost at the same operating voltage and up to 20 percent lower power consumption at equivalent speeds. Additionally, it boasts a 110 percent increase in transistor density, making it exceptionally suited for server applications.
As we look forward to seeing these advancements come to fruition, the developments in next-gen semiconductor processes promise to keep the tech world on its toes. The future of AI capabilities and efficiency may very well hinge on these breakthroughs, making the impending advancements all the more exciting.
