Three developments this week underline how fast the race to build better AI chips is moving — and how many players now want a piece of it.
Intel, the storied American chipmaker that has struggled for years, is showing renewed signs of life in its chip business, according to The New York Times. After a long stretch of setbacks, the report points to the company finding firmer footing in a market increasingly defined by demand for AI hardware.
Meanwhile, OpenAI has unveiled its first custom AI chip designed to power ChatGPT, according to Caliber.Az. Building a bespoke chip is a notable step: companies that run massive AI services often want hardware tuned to their own software, rather than relying solely on off-the-shelf processors from outside suppliers.
And at the cutting edge of manufacturing, IBM has announced a 0.7-nanometer process node and introduced a technology it calls NanoStack, according to More Than Moore. Process nodes are a rough shorthand for how small and densely packed a chip's transistors are; smaller numbers generally signal the ability to pack more computing power into the same space.
Together, the three items sketch a picture of an industry pushing on multiple fronts at once — an established manufacturer regaining momentum, a leading AI company moving to design its own silicon, and a research-driven push toward ever-smaller manufacturing.
Why it matters: AI chips are the foundation that everything from chatbots to scientific tools is built on, so advances in who makes them and how small they can get shape the cost, speed, and availability of the AI products people use.