IBM’s new chip fits roughly 100 billion transistors the size of a fingernail

IBM on Thursday unveiled the first chip made using the latest semiconductor technology, a chip that contains nearly 100 billion transistors in an array of devices no larger than your fingernail. Packing more transistors into a die the same size or smaller than previous generations is a key part of increasing power efficiency and speed.

The size of the new chip is 0.7 nanometers, i.e. smaller than 2nm IBM chip It was first revealed in 2021. But the new chip’s circuitry has been significantly modified. That older, larger process resulted in the transistors being laid out flat in what IBM Research called nanosheets CNET reported in 2021. Now, the new 0.7nm chip uses IBM’s recently developed nanostack architecture, which stacks nanosheets vertically.

IBM says the new architecture results in better performance. In the company’s experiments, it found that the new chip improved performance by up to 50% and energy efficiency by 70% compared to the 2nm version.

According to IBM, the nanostack architecture also allows for a 40% smaller die for SRAM — static RAM is a type of memory that does not require a constant flow of electricity to store data, and because it is faster than DRAM, it is in high demand for AI applications.

However, the chip will not be operational for a while. IBM is still working with its manufacturing partner, Rapidus, a Japanese foundry (chip-making factory), to ramp up projects. IBM says it “sees a path to production” within five years, but demand for energy-efficient computing devices is growing.

Chips like those designed by IBM, Nvidia, AMD and others are the backbone of the AI ​​industry. As AI developers like OpenAI and Google race to build the most advanced models, they require massive amounts of power, or computing, to train them. But this can take a lot of time Electricity, clean water and land To dedicate to data centers.

“Everyone is demanding more performance, but no one wants to pay the energy bill,” said Huiming Bo, vice president of research and development at IBM Semiconductor. He said that the energy efficiency of the new chip is “a critical element of artificial intelligence.”

Creating more efficient devices is a key part of the AI-powered future that technology leaders envision. This has led to the current shortage of production capacity for memory, processors and other components Lack The necessary parts for any laptop or gadget you might want to purchase. New research models may help technology companies — and everyone who uses their products — get more bang for their buck in the future.

“Basically, it boils down to: Can we make transistors more efficient?” said Jay Gambetta, director of IBM Research. “This is a platform that can be customized, so we expect it will impact everything from logic to SRAM, and I expect that as this scales we will see larger, more efficient AI accelerators.”