IBM creates the first sub-1nm chip: a milestone that redefines Moore's Law

IBM has once again marked a before and after in the semiconductor industry. The company announced that it has built the first chip with 0.7 nanometer (7 angstrom) technology, surpassing the 1nm barrier that seemed unattainable. This achievement is not just another number in the race for miniaturization; it represents a fundamental change in transistor architecture and in how we understand Moore's Law. For context, Moore's Law —which predicts that the number of transistors on a chip doubles every two years— has driven technological progress since the 1960s. However, since nodes reached 5nm and 3nm, development costs and physical limitations have slowed the pace. IBM's advance demonstrates that miniaturization still has room to go, but also underscores the growing challenges the industry faces.

What exactly happened?

IBM Research presented a test transistor that uses a stacked nanosheet structure with a 0.7nm gate. Unlike current FinFET transistors, this new architecture allows greater control of current flow and a drastic reduction in energy consumption. The chip was fabricated at the Albany Nanotech Research Center in New York and, although it is a laboratory prototype, it demonstrates the viability of the technology. According to The Next Web source, the transistor uses a stacked nanosheet structure that improves performance and efficiency. It is important to note that the 0.7nm node does not correspond to an exact physical dimension of the gate, but rather is a commercial designation indicating a transistor density equivalent to what would be expected at that node. This type of nomenclature has become common in the industry: for example, TSMC's 5nm nodes do not have 5nm dimensions, but represent an improvement over the previous node.

Why is it important?

The advance is crucial for several reasons. First, because the industry had been stagnant at 3nm and 5nm for years, with difficulties scaling further. Second, because a sub-1nm transistor allows packing more transistors in the same space, which translates into up to 50% higher performance and energy efficiency that could double the battery life of mobile devices. Third, because this milestone comes at a key time for artificial intelligence, where computing demand is growing exponentially. To put it in perspective, Apple's current 5nm chips (like the M1) already offer remarkable performance, but a leap to 0.7nm could multiply processing power in mobile devices, allowing AI models with billions of parameters to run locally without relying on the cloud. Additionally, in data centers, reducing energy consumption is critical: according to data from the International Energy Agency, data centers consume about 1% of the world's electricity, and this figure is expected to increase with the expansion of AI. More efficient chips could mitigate that impact.

Consequences for the industry

IBM's announcement does not mean we will see 0.7nm chips on the market tomorrow. It is a proof of concept that could take years to reach commercial production. However, it sends a clear signal to competitors (TSMC, Samsung, Intel) that miniaturization still has a way to go. It also forces a rethink of investments in extreme ultraviolet (EUV) lithography and new materials like silicon-germanium. Historically, IBM has been a pioneer in semiconductor innovations: in 2015 it demonstrated a 7nm transistor, and in 2021 it presented a 2nm chip. But unlike then, the competitive landscape is now more complex. TSMC and Samsung are already mass-producing 3nm nodes, and Intel plans to launch its 1.8nm (20A) node in 2024. IBM's achievement could accelerate the roadmap of these manufacturers, but mass production at 0.7nm will require advances in high numerical aperture EUV lithography (High-NA EUV), which is still under development. According to ASML, the leading supplier of EUV machines, the first High-NA tools will be delivered in 2025, but volume production is not expected until 2027-2028. Therefore, it is reasonable to project that the first products based on this technology will arrive around 2028-2030, as the source indicates.

What readers should know

Do not confuse the node name with actual physical dimensions. The '0.7nm' is a commercial designation that does not exactly correspond to the gate size. Additionally, mass production of these chips will require advances in lithography and manufacturing processes that are not yet ready. The first products based on this technology are expected to arrive around 2028-2030. It is important to note that IBM does not manufacture chips at scale; the company typically licenses its technology to other manufacturers or collaborates with them. For example, its 7nm node was used by GlobalFoundries, although it later dropped out of the race for advanced nodes. In this case, IBM could partner with Samsung or Intel to bring the technology to market. Also consider the cost: advanced node chips are extremely expensive to design and manufacture. A 3nm design can cost over $500 million, according to IBS. Therefore, the 0.7nm chip will only be viable for high-value applications such as AI servers, supercomputing, and high-end devices.

“This is an impressive engineering achievement, but we must not forget that Moore's Law is not just about size, but about cost and performance. IBM has shown it is possible, but the real challenge is making it profitable.”

Impact on the future of work and AI

For TheVortiq readers, this advance has direct implications for automation and artificial intelligence. More powerful and efficient chips will allow running more complex AI models on local devices, reducing reliance on the cloud. They will also improve productivity in data processing and simulation tasks. Software and SaaS startups must prepare for hardware that will make applications unthinkable today possible. For example, inference of large language models (LLMs) like GPT-4 could be performed on a smartphone, opening the door to much more capable personal assistants without network latency. In the workplace, AI-based automation tools could run complex analyses in real time, transforming sectors such as medicine (diagnostic imaging), logistics (route optimization), and engineering (simulations). Additionally, energy efficiency is key for sustainability: according to a study by the University of Massachusetts, training an AI model can emit up to 284 tons of CO2, equivalent to five times the emissions of an average car over its lifetime. Reducing chip energy consumption would help mitigate this impact. In summary, IBM's advance is not only a technical milestone but a catalyst for the next wave of innovation in AI and automation, which will redefine the future of work.