
TSMC's new 1.4nm method promises faster, more power-efficient chips
What's the story
Taiwan Semiconductor Manufacturing Company (TSMC) has unveiled a game-changing tech to make chips faster.
The new A14 method (1.4nm class) promises processors that are either 15% faster at the same power consumption as its N2 (2nm process) chips or consume 30% less power at the same speed.
The company will launch A14 technology in 2028. Going forward, high-performance (A14X) and cost-optimized (A14C) versions may launch after 2029.
Chip integration
TSMC unveils new chip design system
Along with its A14 manufacturing tech, TSMC has also unveiled plans for a new chip design system, "System on Wafer-X."
This is an innovative approach that will combine at least 16 large computing chips with memory chips and fast optical interconnections.
The system will also feature cutting-edge tech that can deliver thousands of watts of power directly to the chips.
Tech comparison
'System on Wafer-X' to surpass NVIDIA's multi-chip approach
TSMC's upcoming "System on Wafer-X" will surpass NVIDIA's existing flagship graphics processing units (GPUs) which are made up of two large chips stitched together.
The tech will even be better than NVIDIA's "Rubin Ultra" GPUs, due in 2027, which will combine four chips.
This development highlights TSMC's dedication to advancing the limits of chip design and integration.
Industry rivalry
TSMC's expansion plans coincide with Intel's competitive strategy
To back its new technologies, TSMC plans to build two factories close to its chip plants in Arizona, US.
The development comes as Intel, another giant in the industry, gears up to unveil new manufacturing technologies next week.
Intel has ambitions to overtake TSMC in making the world's fastest chips. The increasing demand for AI chips has turned the race between the two from speed-focused production to complex chip integration.