Do these 3 things before closing this tab:
1Repair Windows errors before they cause bigger problems2Fix the driver behind crashes, sound loss and screen glitches3Clear out junk files and repair common Windows errorsIntel announced Pohoiki Springs on March 18, 2020, as a rack-mounted research computer with computational capacity equivalent to 100 million neurons. It combined 768 Loihi neuromorphic chips and was made available to members of Intel’s Neuromorphic Research Community (INRC) through cloud access—not sold as a consumer computer. Intel’s announcement presented it as a platform for studying brain-inspired computing, not as a machine with animal-like intelligence.
What was Pohoiki Springs?
Pohoiki Springs was a data-center system built to help researchers investigate neuromorphic computing. Intel said its 768 Loihi research chips were housed in a rack-mounted chassis about the size of five standard servers, with a stated capacity equivalent to 100 million neurons. Intel’s Mike Davies described its power level as under 500 watts in the 2020 announcement.
Access was intended for INRC members via Intel’s cloud environment. Researchers could develop applications with the Nx SDK and community-contributed software components. Intel said the system was meant to support work on real-time processing, problem solving, adaptation and learning. The company also emphasized that its neuromorphic systems were still in the research phase and were not intended to replace conventional computers.
What does “neuromorphic computer” mean?
A neuromorphic computer uses hardware and computing methods inspired by nervous systems. In Intel’s description, the relevant ideas include parallel processing and asynchronous signaling. Rather than continually processing every possible input in the same way, event-driven spiking neural networks can respond to activity as it occurs, with the aim of limiting unnecessary computation and data movement.
Recommended Free Tools
#1 Best Overall
Intel’s later explanation of Loihi 2 also describes sparse, event-driven computation, asynchronous spiking networks, and memory integrated with computation. These are architectural approaches for particular workloads; they do not make a neuromorphic processor a digital copy of a biological brain. Intel’s neuromorphic computing overview provides its current research context.
What could researchers use it to study?
Intel cited several workload areas its research partners were exploring. They illustrate the system’s intended use, not a promise that it would outperform conventional processors across the board.
- Constraint satisfaction: scheduling and package-delivery planning.
- Graph and pattern searches: shortest-path identification and approximate image search.
- Optimization: allocating bandwidth across wireless channels.
Intel’s 2020 announcement said Loihi could process certain demanding workloads up to 1,000 times faster and 10,000 times more efficiently than conventional processors. Those were Intel’s claims for certain workloads, not universal benchmarks or guarantees for every application.
Does 100 million neurons mean it was as capable as an animal brain?
No. The figure described computational capacity in Intel’s system; it is not a measure of general intelligence, awareness, or broad cognitive ability. A hardware neuron count does not establish that a computer can think or behave like an animal with a similar number of neurons. Intel used the comparison to convey scale while describing an experimental architecture for researchers.
Rank #3
How did Pohoiki Springs compare with Hala Point?
Hala Point is a later Intel research prototype, announced on April 17, 2024, with a different processor generation and a larger stated neuron capacity. Intel said it was initially deployed at Sandia National Laboratories. The figures below are specifications reported by Intel, not a direct measure of which system performs better on any particular task.
| System | Announcement and status | Neuron capacity | Processors | Power specification | Access or deployment |
|---|---|---|---|---|---|
| Pohoiki Springs | Announced in 2020 as a research system | Equivalent to 100 million neurons | 768 Loihi research chips | Under 500 watts, as described by Intel’s Mike Davies | Cloud access for INRC members |
| Hala Point | Announced in 2024 as a research prototype | Up to 1.15 billion neurons | 1,152 Loihi 2 processors | Maximum 2,600 watts | Initially deployed at Sandia National Laboratories |
Intel also reported 128 billion synapses and 140,544 neuromorphic cores for Hala Point. Its release tied some performance figures to characterized workloads. For example, the reported efficiency of up to 15 TOPS/W for a deep neural network applied to a specific 8-bit multilayer perceptron configuration with pruning and activation conditions; Intel noted that results may vary. It should not be treated as a general efficiency guarantee. See Intel’s Hala Point announcement for its specifications and qualifications.
Independent reader supportYour contribution helps us test, update, and keep practical guides available for everyone.Can you buy Pohoiki Springs or Loihi chips?
The cited Intel materials describe Pohoiki Springs and Loihi as research hardware and do not establish consumer purchase availability. The access route Intel described for Pohoiki Springs was cloud-based participation in the INRC, rather than buying the rack system. Intel’s current research pages describe Loihi 2, the Lava software framework, Kapoho Point—an eight-chip Loihi 2 board—and the INRC. Intel says INRC membership is free and open to qualified groups; this is a research collaboration route, not evidence that the hardware is a retail product. Details are available in Intel’s neuromorphic research overview and its Loihi 2 announcement.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
