NVIDIA’s GPUs are already powering some of the most advanced machine learning systems for technology companies – and the demand is only increasing.

Google, Amazon, and Facebook are just a few companies investing heavily in machine learning – the branch of AI that allows the tech companies’ computers to learn information on their own that they weren’t programmed to know.

Google, for example, uses its own TensorFlow machine learning systems for its Google Translate speech recognition app, Google Photos, Gmail, and its Web searches. And as these companies dive further into machine learning, they’re building their own complex computers using graphics processing units (GPUs) to power them – and that could be particularly beneficial for NVIDIA. NVIDIA makes some of the most popular GPUs for gaming, but the hardware is increasingly finding its way into supercomputers.

Facebook already uses NVIDIA’s Tesla M40 GPU accelerators to help power its Big Sur machine learning computers. These NVIDIA GPUs were specifically designed to train deep neural networks for enterprise data centres, and the company says they’re 10 to 20 times faster than other neural network computers. NVIDIA says its GPU-powered machine learning computers can help train neural networks to learn new things in just a few hours, as opposed to days or weeks with less powerful systems.


The Real Potential

Many of the world’s best technology companies are already using NVIDIA’s GPU-powered machine learning systems, but how much could NVIDIA make from this? According to MarketsandMarkets, the cognitive computing market (which includes natural language learning, machine learning, and automated reasoning) is expected to be worth $12 billion by 2019. And NVIDIA’s already positioned itself to benefit from the enterprise machine learning market. Back in April, the company released its new Tesla P100 GPU for corporate data centres.



NVIDIA Tesla P100 Accelerator

The company said these chips pack 15 billion transistors on them, which is about twice as many as Intel’s recently debuted server processors.
This is important to note because technology companies are increasingly looking to advanced cloud computing systems. The global cloud computing market is currently worth $204 billion, and NVIDIA is already tapping into this market by supplying cloud computing GPUs to Microsoft, Google, and Amazon.



NVIDIA’s bread and butter is still its gaming revenue, which comprised $687 million of the company’s $1.3 billion total revenue in fiscal Q1 2017. Meanwhile, its data centre revenue, which includes its GPU sales for cloud-based and machine learning services brought in just $143 million in the quarter. That said, the company has recently started focusing on segments outside of gaming, and it has made huge moves in a short amount of time. The company’s data centre revenue increased by 63% year over year in the last quarter, for instance.

Investors looking for NVIDIA machine learning growth should keep a close watch on the company’s data centre revenue, particularly next year when the Tesla P100 starts finding its way into more servers. With the company’s growing list of machine learning GPU customers, NVIDIA is poised to benefit from machine learning’s growth (and its thirst for more complex data processing). But investors are going to have to wait at least a few quarters to find out how well these machine learning investments have paid off.


For the latest news and conversations about AI in business, follow us on Twitter @Business_AI and join us on LinkedIn – AI Business Community


Embedded image credit: http://nvidianews.nvidia.com/multimedia

Source: http://bit.ly/1XW59rx