IBM Risks $3 Billion to Stay Relevant in Chip Biz

2 minute read

IBM says it’ll drop $3 billion into a five-year bucket designed to help it shore up its research and early stage development programs. The idea, says IBM, is to generate new chip-related technologies that can power evolving cloud computing and “Big Data” systems.

The move comes as confidence in Moore’s Law — the more rule-of-thumb-than-law that says the number of transistors you can stick on a computer chip will double every two years, resulting in periodic increases in computing power — has been dwindling.

The money IBM’s spending will focus on several confidence-bolstering programs, the first of which targets “7 nanometer and beyond” silicon tech and the basic physics-related challenges governing size and production: the smaller silicon-based chips get, the harder they are to manufacture, and we’re currently approaching a physics-related size wall.

“The question is not if we will introduce 7 nanometer technology into manufacturing, but rather how, when, and at what cost?” IBM Research Senior VP John Kelly said in a press release, calling scaling to 7 nanometers or smaller “a terrific challenge.”

And that’s where IBM’s second research pole comes in, says the company, aimed at coming up with alternative ways to think about how chips are made in a post-silicon world. Think about Intel’s 22-nanometer 3-D tri-gate processors — still silicon-based, but employing a relatively radical design shift that allowed it to improve switching states dramatically and consume half the power of older 32-nanometer chips.

But as IBM notes, dropping to below 7 nanometers would require “a new kind of material to power the systems of the future.” Those alternatives? IBM lists carbon nanotubes (specialized cylindrical nano-structures), graphene (a pure form of carbon potentially superior to silicon), and III-V technologies (metal-oxide as opposed to silicon semiconductors), as well as neuromorphic computing, neurosynaptic computing and cognitive computing (systems that mimic the human brain or nervous system), silicon photonics (moving data with light), machine learning techniques (artificial intelligence) and of course, quantum computing.

IBM says its research teams will combine scientists and engineers from Albany and Yorktown, New York; Almaden, California and others in Europe. And to be clear, it says these aren’t brand new research areas — this is IBM channeling cash to existing programs that are already underway. In other words, it’s a confidence-bolstering move involving the B-word as much as anything.

More Must-Reads From TIME

Write to Matt Peckham at matt.peckham@time.com