IBM has unveiled the world’s first sub-1-nanometer chip technology – a 0.7nm (7-angstrom) design it calls “Nanostack” that packs roughly 100 billion transistors onto a fingernail-sized chip and runs about 70% more efficiently than today’s 2nm technology. It is a genuine engineering milestone. It is also years away from anything you can actually buy. Here is what was really announced, what “sub-1nm” actually means, and why it matters.
On June 25, 2026, IBM Research said it had pushed chip technology below the one-nanometer barrier for the first time, into what the industry now calls the “angstrom era” – the point where transistor features approach the size of individual atoms. The headline figure, 0.7nm, sounds almost impossibly small, so let us unpack it honestly rather than just repeat the press release.
What IBM actually announced
The breakthrough is a new transistor architecture IBM calls Nanostack, demonstrated at the 0.7nm (7-angstrom) node. In a chip the size of a fingernail, that works out to roughly 100 billion transistors. IBM says the design is about 70% more power-efficient, or up to 50% more powerful, than the 2nm node it first showed back in 2021. Crucially, this is a research demonstration of the underlying technology, not a product rolling off a production line.
What “0.7nm” really means (it is not a ruler)
Here is the reality check almost nobody mentions: modern node names like “0.7nm,” “2nm,” or “3nm” are not literal measurements. No part of the chip is actually 0.7 nanometers wide. Years ago the number roughly described a physical feature, but today it is essentially a marketing label for a generation of technology – a shorthand for “denser and more efficient than the last one.” So when you read “0.7nm,” read it as “the next big step in squeezing more transistors into the same space,” not as a tape-measure reading.
Nanostack: stacking transistors in 3D
The clever part is how IBM gets there. Nanostack builds on nanosheet technology – the current leading-edge approach, which IBM also pioneered – by vertically stacking and staggering transistors using 3D sequential integration. Instead of only spreading transistors out across a flat surface, it builds upward, packing far more into the same footprint. The architecture also allows different material combinations in each stacked layer, so each transistor can be tuned for performance or efficiency independently. Think of it as going from a single-storey building to a smartly designed tower.
Why it matters, especially for AI
Two things make this a big deal: density and efficiency. More transistors in less space, drawing less power, is exactly what modern computing – and AI in particular – is starving for. IBM’s researchers estimate that an AI accelerator built on 7-angstrom technology could deliver around 9,000 TOPS (trillions of operations per second), roughly six times the ~1,500 TOPS of today’s popular AI chips. In plain terms, that points toward AI that is faster, cheaper to run, and far less power-hungry – which matters for everything from data centres to the AI agents starting to appear in everyday software. It is also a strong signal that Moore’s law, repeatedly declared dead, still has room to run.
The reality check: this is research, not a product
This is where honesty matters. A lab milestone is not a shipping chip. By IBM’s own roadmap, widescale adoption of 2nm devices is only expected toward the end of the decade, followed by 1.4nm and 1nm nodes after that – and sub-1nm production comes later still. Turning a research demonstration into something that can be manufactured by the billion, reliably and affordably, is the genuinely hard part, and it takes years. So this is a glimpse of the floor we have not hit yet, not a chip arriving in next year’s laptop.
What it means for you
Nothing you need to act on today, but plenty to be encouraged by. The takeaway is that the steady march of more powerful, more efficient hardware is not stopping – the devices and AI tools you use will keep getting better for years to come. If you are weighing a hardware purchase now, do not wait for 0.7nm; buy what fits your needs today, and know the upgrades after that will be even better. Our guide on the signs it is time to upgrade your GPU is a more useful read for the here and now. For the full technical announcement, see IBM Research’s own write-up.
What the ‘angstrom era’ really signals
The shift to angstrom-scale naming is more than a marketing reset – it marks a genuine change in how chips advance. For decades, progress came mostly from shrinking transistors flatter and smaller. We are now close enough to the size of individual atoms that simply shrinking further runs into the hard limits of physics. So the industry is changing tactics: building upward and stacking in three dimensions, as Nanostack does, and mixing new materials rather than relying on shrink alone. In other words, the headline number keeps falling, but the real story is architecture – clever design squeezing more out of roughly the same physical space. That is exactly why many engineers are more optimistic about the next decade of computing than the steady drumbeat of “Moore’s law is dead” headlines would have you believe.
Frequently asked questions
Is a 0.7nm chip actually 0.7 nanometers?
No. Modern node names are technology-generation labels, not physical measurements. Nothing on the chip is literally 0.7nm; the name signals a new, denser, more efficient generation.
When will I be able to buy a sub-1nm chip?
Not soon. IBM’s roadmap puts widescale 2nm adoption near the end of the decade, with 1.4nm and 1nm after that, and sub-1nm later still. This is a research milestone, so realistically you are looking years out.
Does this mean Moore’s law is not dead?
It suggests the slowdown has been exaggerated. By stacking transistors in 3D rather than only shrinking them, IBM shows there is still meaningful headroom for packing in more compute.
Is IBM the only company doing this?
No. Intel, TSMC, and Samsung are all racing toward angstrom-era nodes with their own roadmaps. IBM is a research powerhouse whose work often filters into the wider industry, but it is one player in a very competitive field.
Bottom line: IBM’s sub-1nm Nanostack is a real and impressive milestone that says the future of computing still has plenty of room to grow – just do not expect a 0.7nm chip in your hands any time this decade.
Jared is a tech journalist covering product launches, industry news, and the culture around technology. He has been reporting on the consumer tech beat for more than eight years.