IBM unveils 0.7-nanometer chip technology with a new 3D transistor design, aiming to deliver faster AI performance and greater energy efficiency while pushing semiconductor innovation beyond the 1nm barrier.
 |
| IBM's latest chip breakthrough could shape the next generation of AI computing by combining higher performance with lower energy consumption, although commercial production is still years away. Image: CH |
Tech Desk — June 26, 2026:
IBM has taken another step in the race to build more powerful chips for artificial intelligence, unveiling what it says is the world's first technology capable of producing semiconductors smaller than one nanometer.
The company introduced a 0.7-nanometer chip technology built around a new three-dimensional transistor architecture called Nanostack. IBM says the design allows nearly 100 billion transistors to fit onto a chip about the size of a fingernail, dramatically increasing computing density.
According to IBM, the new technology can deliver up to 50% higher performance than its 2-nanometer chip announced in 2021. Alternatively, it can achieve the same level of performance while using up to 70% less energy, a key advantage as AI systems continue to demand more computing power.
The company also says it has reduced the size of SRAM memory cells by 40%. That improvement could help AI processors move data more efficiently, reducing bottlenecks in both training and inference workloads.
The announcement comes as the semiconductor industry faces growing pressure to keep advancing chip performance despite the increasing difficulty of shrinking transistor sizes. For years, manufacturers have relied on Moore's Law to deliver more computing power by fitting more transistors onto a single chip. As physical limits become more challenging, new transistor designs and manufacturing techniques are becoming just as important as making chips smaller.
IBM's latest research reflects that shift. Rather than focusing only on size, the company is also emphasizing smarter chip architecture and improved energy efficiency—two factors that are becoming critical as AI applications expand across cloud computing, data centers, robotics, and consumer devices.
The potential benefits extend beyond faster AI models. More efficient chips could help reduce electricity consumption in large data centers, lowering operating costs while supporting increasingly complex AI workloads. That balance between performance and power use is expected to play a major role in the next generation of computing.
Still, the technology is not ready for commercial production. IBM estimates it could take around five years before the process reaches manufacturing, and the company has not yet announced a production partner. Moving from a laboratory breakthrough to mass production will require overcoming significant engineering and fabrication challenges.
Even so, the announcement signals that competition in advanced semiconductor technology remains intense. As demand for AI computing accelerates worldwide, breakthroughs in chip design are likely to become a defining factor in determining which companies lead the next era of artificial intelligence.