Tech Marketplace is a series of edited insights from leaders in business, technology, and academia on technology’s role in growing businesses. Find more stories here.
Microsoft has embraced a technology that could deliver incredibly low power computing. The tech — dubbed the Neural Network Processor — could significantly cut power consumption by using parallel computing and graphics processing, especially on data-intensive tasks.
The technology is not available at the hardware level yet, but the processor is the next step in the company’s bid to build supercomputer capable of computing in the cloud. One of the tech company’s big bets is on making high powered data centers easier to build and run, and getting more of the power of supercomputers onto the same computing devices that consumers use.
The chip uses technologies from hardware that Microsoft already sells — like the graphics processing units in the Xbox. And, it’s also built into software that will arrive over the next few years.
This type of technology makes the company’s cloud computing division much more valuable. While Microsoft’s cloud computing division already holds 21 percent of the market, some analysts estimate that it could have a market share as high as 40 percent in five years.
Microsoft declined to share specific performance numbers and specifications of the chip, which is still under development. However, some of the technology it could deliver is the result of a research project that lasted more than a decade.
For more about the chip and Microsoft’s other data-driven initiatives, check out “Project Natal,” the technology powering Microsoft’s new augmented reality system called “HoloLens,” and “AI Powered Game Audio,” which was part of the company’s much publicized augmented reality experiment.