Analysts, however, are now worried about profit margins, as gains from the company's market share come mostly from (price) hanging fruits. But investors should note that the company is targeting the upper end of the market with its next-generation GPU platform - the Vega.
Why AMD wants to go high-end?
Internal company estimates suggest that the high-end GPU market represents less than 15% of sales, but represents two-thirds of margins. The company failed to penetrate this segment with the Polaris GPUs, which are mainly aimed at mid-range desktop computers and below.
Therefore, the company is losing in the most lucrative segment of the GPU market, but believes that the Vega will help address this opportunity.
Bringing the game to NVIDIA
NVIDIA dominates the high-end GPU market, but AMD believes that its Vega GPUs are capable enough to compete against its rival's premium offerings. As a result, test results from the Baidu DeepBench Open Source benchmark indicate that the Vega can outperform the NVIDIA P100 Tesla GPU in auto-learning applications by a fair margin.
Now, NVIDIA claims that the Tesla P100 is one of its most advanced GPUs, with specifications to substantially increase the computing power of a data center, while reducing costs by 60% simultaneously. NVIDIA's Tesla GPUs have already made significant strides in the cloud computing market, with users like Microsoft and Tencent using them to advance their artificial intelligence capabilities.
But AMD can make a dent in this market if Vega manages to surpass NVIDIA's first-line offer in independent machine intelligence testing. In addition, AMD leaves nothing to chance as it has launched a number of open source software solutions for developers to take advantage of their GPU accelerators.
The company's Radeon Open Compute platform is designed to support a variety of AI frameworks including Facebook's Caffe and Google's TensorFlow machine learning library, among others. As such, it will not be surprising that the cloud computing GPU market plays a critical role in driving AMD's results, especially since cloud data center workloads could grow 26% annually up 2020.
The possible financial impact
AMD believes that its expansion in high-end markets such as cloud computing and machine intelligence will increase profit over time. The company expects gross margin to reach 36% in 2018, up from 31% last year. In addition, AMD expects further expansion of its margins - up to 40% or more - by 2020 due to improved product mix.
Of course, whether or not the company will enjoy this increased profitability depends on how their new products work in the real world. But as something to keep investors optimistic, AMD estimates that the data center market alone will be a $ 21 billion opportunity in just a few years.
Investors will have to wait until the end of this year before the launch of the new Vega GPU. If AMD can offer you an attractive offer to tackle the next generation of cloud computing, your top and bottom lines will get you shot in the arm by capturing only a small portion of this thriving market.
