Intel: Reports of Moore's Law's death have been greatly exaggerated

But the future of computing may look very different from today's smartphones.


Despite reports throughout the years declaring the death of Moore’s Law, it’s still alive and well, says an executive at chipmaker Intel.

Jim Keller, Senior Vice President of Intel’s Silicon Engineering Group said Tuesday at MIT Technology Review’s Future Compute conference that Moore’s Law — a refresher definition is below — won’t be dead for at least another 30 years or so. (You might say the famous Twain quotation applies.)

Moore’s Law was originally described by Intel engineer and co-founder, Gordon Moore, in 1965 and claimed that “the number of transistors incorporated in a chip will approximately double every 24 months.” In other words, every two years, the physical hardware of computing will become smaller and more capable, thus enabling advances in computational power as well. Since 1965, this technological growth estimate has brought us from large computing mainframes to sophisticated smartphones.

Thus, the “death” of Moore’s Law has the potential to lead to the stagnation of technology itself, but Keller suggests that this perspective fails to acknowledge much of the complexity of Moore’s Law.

"Moore’s Law was never one thing

“If you go look at Moore’s Law, Moore’s Law was never one thing,” Keller told the assembled audience. “There was transistor architecture, strain, materials, 3D architecture… Every single component of shrinking the transistor changed year over year. There’s been an endless stock of them.

“Imagine a business where a million people wake up every day, working on making things smaller and better. It was that collective belief system about this [that helped] the inventions keep coming.”

MIT Professor, Max Shulaker, presents his team's work in building carbon nanotube-based microprocessors during day one of MIT Technology Review's Future Compute conference in Cambridge, Massachusetts.

Sarah Wells

Just as smartphones were unimaginable to generations before us, Keller suggests that future innovation in computing may look like nothing we’ve seen today. And these changes are already happening, as MIT Professor of Electrical Engineering and Computer Science, Max Shulaker, said later that same morning at Future Compute.

“If we really want to revolutionize computing [and] not just receive a 10 or 20 percent benefit but a 100 or a 1,000 x benefit,” said Shulaker, “then business as usual is insufficient. And new innovations are required.”

From Shulaker’s perspective, the path forward might not necessarily be to continue shrinking the current silicon-based chips standard in industry, but instead to build taller, more computationally powerful chips, based on something else entirely: carbon nanotubes.

While only discovered in the early ‘90s, the flexible applications of these incredibly small carbon nanotubes have made them useful for everything from medical drug delivery to hardware development. Particularly interesting to Shulaker and his research team, as well as the future of computing at large, is how these nanotubes could be used to replace silicon transistors in a microprocessing chip.

An audience waits for speakers to take the stage on December 2, 2019 at MIT Technology Review's Future Compute conference in Cambridge, Massachusetts.

Sarah Wells

Standard silicon processing chips in electronics are built in parallel, Shulaker explained to the audience at Future Compute, meaning that information between chips has to go through a tedious process in order to send and receive information. The advantage of carbon nanotubes in this context, said Shulaker, is that instead of different processors being built in parallel they could be built directly on top of one another — dramatically improving data transmission speeds.

The 3D model Shulaker showed to demonstrate this concept looked much like a parking garage, with different vertical microprocessing layers being supported by carbon nanotube transistors.

While some of Shulaker’s carbon nanotube designs have yet to be fully realized, he did demonstrate to the audience that a working model was already been fabricated by his team earlier this year.

To demonstrate the processor’s success, Shulaker shared with the audience the processor’s (and all computer’s) first words, “Hello, World! I am RV16XNano, made from CNTs.”

Whether or not this is a natural extension of Moore’s Law, or something else entirely, is still up for debate. But Keller emphasized at Future Compute that we have a lot more to lose from not believing in Moore’s Law than we do if we continue to believe in it.

“If Moore’s Law is alive, and you don’t think it is, you’re not going to do the things necessary to deal with the complexities driven by Moore’s Law. Things will get complicated and out of control. So your stance on this question is super important [when] driving an engineering business or any kind of computing business. Moore’s Law is literally relentless.”

Nevertheless, both Keller and Shulaker agreed that Silicon Valley wouldn’t be renamed “Carbon Nanotube Valley” anytime soon.