Neuromorphic & Alternative Compute
Maturity
2 / 5
Basis, source, vintage
Research and pilot deployment only; the constraint is the surrounding software ecosystem, not the devices.
Source: Intel Hala Point (announced 17 Apr 2024, deployed at Sandia National Laboratories): 1,152 Loihi 2 processors, 1.15bn neurons, up to 20 peta-ops/s and 15 TOPS/W at 8-bit in 10:1 sparse networks, max 2,600 W. Intel's Mike Davies: 'not yet close to commercialization' (2024)
What it is
Processor architectures that do not follow the conventional model — neuromorphic chips, analogue and in-memory computing, photonic processing.
What it does
It addresses the constraint that conventional architectures are hitting: energy per operation. If AI's growth is bounded by power rather than by silicon, the answer is architectural rather than incremental.
Common misconception
That this is a niche research topic. It is the most plausible route around an energy ceiling that is already shaping where data centres can be built.
Steepens
- Automation of Cognitive Work — moderate · ○ shared candidate constraint (with 4 others)
- Energy Transition and Electrification — moderate
- Precision Everything — marginal · marginal: not counted in the minimum
What would change its gradient
- A software toolchain that does not require rewriting everything
- Energy cost per operation becoming the binding commercial constraint, which would force the switch
- Fabrication access, currently concentrated in very few places
Shares megatrends with
- Adoption Appetite
- AI & Machine Learning
- Autonomous Vehicles
- Capital Availability
- Cloud & Datacentre Infrastructure
- Data & Datafication
In the book
Why the Future Arrives Unevenly — Ch. 8 — The Rebuild of the Physical Base · Ch. 14 — How Hype Is Manufactured
Sources
- Research and pilot deployment only; the constraint is the surrounding software ecosystem, not the devices. Source: Intel Hala Point (announced 17 Apr 2024, deployed at Sandia National Laboratories): 1,152 Loihi 2 processors, 1.15bn neurons, up to 20 peta-ops/s and 15 TOPS/W at 8-bit in 10:1 sparse networks, max 2,600 W. Intel's Mike Davies: 'not yet close to commercialization' (2024).