Advanced Micro Devices, Inc. Outlines Its Machine Learning Strategy
- Advanced Micro Devices, Inc. recently announced its Machine learning strategy by unveiling the Radeon Instinct accelerators.
- The AI market is expected to reach $36B in less than a decade.
- Can AMD take a significant chunk of this pie? If yes, this could be a key growth driver for AMD Stock?
Sunnyvale, California-based AMD (NSDQ:AMD) recently unveiled their machine learning strategy with the launch of the new Radeon Instinct accelerators, MIOpen library and the ROCm software (launched earlier). Artificial intelligence (AI), Machine Learning (ML) and Deep Learning(DL) have become the new buzzwords in the tech space. Most major technology companies including Alphabet (NSDQ:GOOGL), Microsoft (NSDQ:MSFT), International Business Machines (NYSE:IBM), Facebook (NSDQ:FB) have taken the AI plunge, so it's no surprise that AMD unveiled their machine learning strategy. It was only a matter of time. So how exactly does this alter AMD's prospects? And what does it mean for AMD stock investors?
AI Market Is Expected To Explode In The Next Decade
As per a Tractica research report cited by TOP500 News, the AI market is expected to reach a market size of $36.8B by 2025. To put things into perspective, the firm estimates the current market size at $643.7M (2016), which implies a 57x increase, or an average annual growth rate of over 56% through 2025. And, this is a market which is only expected to increase going forward. Of the total AI market, Tractica estimates that the Deep learning market is worth $304.8M today and could grow to $16B by 2025, representing the largest slice of the AI pie. (See also: Why AMD Stock Looks Unstoppable For Now: Advanced Micro Devices Inc)
Can AMD Take A Chunk Of The Growing AI Market?
This is a market which AMD has till now, ceded to Nvidia. As per the Top500 News report referenced above, Nvidia has attributed nearly half of its data center revenue to deep learning. Based on NVIDIA's first 3 quarter results, the company has racked up over $260M in deep learning revenue ($534M total Data center revenue), accounting for over 80% of the deep learning market. Well, AMD's new machine learning strategy will certainly aim to change this. While AMD is certainly late to the party (NVIDIA is already generating hundreds of Millions of Dollars in AI revenue, growing at an exponential rate), the size of the opportunity ahead means its 'better late than never' for AMD.
How Does The Radeon Instinct Range Stand Up Against Competition?
AMD has launched three versions of its Radeon Instinct range of accelerators, one based on each its Polaris, Fiji and upcoming Vega GPU architectures. The MI6 (based on the Polaris architecture) delivers 5.7 teraflops (TFLOPS) of peak FP16 performance at a power draw of under 150W and 16GB of memory. The MI8 based on the highly energy efficient Fiji GPU delivers 8.2 Teraflops of peak FP16 performance at a power draw of under 175W and 4 GB of High Bandwidth Memory (HBM). The MI25 will be based on the upcoming VEGA GPU architecture and will have a power draw of less than 300W. While no TFLOPS numbers were spilled, the naming convention (MI for Machine Intelligence and the number indicating TFLOPS) suggests the MI25 accelerator could deliver around 25 TFLOPS of peak FP16 performance with 4/8GB of HBM memory. While the MI6 and MI8 are inference accelerators, the MI125 is aimed at the deep learning training market. So, how do these fare against the competition? (See also: AMD Stock Vs INTC Stock: Better 2017 Pick - Advanced Micro Devices Inc Or Intel Corporation?)
As per a Top500News report:
The MI25 will compete head-to-head with the P100 Tesla, NVIDIA’s flagship GPU for training neural networks. It too maxes out at 300 watts. But at 21 FP16 teraflops, the P100 delivers somewhat less peak performance than its new cousin at AMD.
Although the peak flops metrics are relatively close, AMD showed a chart in which both the MI8 and MI25 were beating NVIDIA’s Titan-X GPU on the DeepBench GEMM test, a benchmark used to approximate performance on deep learning types of codes. In fact, the MI25 card outperformed the Titan-X by about 50 percent, despite being only about 20 percent faster in peak performance. The Titan-X uses the same chip as the higher end P100, but lacks the high bandwidth memory of its Tesla counterpart. Regardless if it was the memory, the specific benchmark implementation, or some other factor that contributed to the quicker times on the MI25, performance on real life applications will be the real test for the new GPU
While the AMD range is yet to be put through real world testing, it is safe to conclude that AMD has certainly made a big splash in the AI space with a range of products which certainly stand up to the best in the competition (as far as the benchmark tests indicate).
What Does It Mean For AMD Stock Investors?
A 15% slice of the rapidly growing AI deep learning market could equate to incremental revenue of $2.4B in 2025. To put things in perspective, AMD reported revenue of $4.12B over the last 12 months and $3.99B in 2015. This is down from $5.65B in 2006. With its fortunes hinging on the lower end of the GPU market (usually selling lower performance products at a discount to competitors) and the wild swings in the CPU space, the machine learning strategy could provide AMD with stable growth, something the company hasn't been successful with. AMD is certainly off to a great start considering that the recently unveiled products match up to the best from GPU rival NVIDIA, but this is just a beginning. The company has a long way to go in order to justify the faith investors have put in the stock. AMD stock price is up 267% in the year-to-date.
Putting It All Together
AMD recently launched its Radeon Instinct line of accelerators, which are at the center of the company's machine intelligence strategy. While its still too early to decide, the potential of the market means a huge opportunity beckons AMD. Any significant market share gains could go a long way in improving the fundamentals of the company. If you have the risk appetite, AMD stock is a great way to play the growth of the AI market. For risk-averse investors, there is always the safer option to buy NVDA stock, which leads the deep learning market. All in all, AMD is certainly off to a great start with its machine learning initiatives and AMD stock investors are in for an exciting time.
Looking for great tech stocks? Check out our top technology stock picks, which have beaten the NASDAQ by over 110%.