Qualcomm (NASDAQ:QCOM) stock has been on a smartphone roller coaster, and that ride has been bumpy lately, with smartphone sales declining while a legal battle with Apple (NASDAQ:AAPL) rages on. Diversification into laptop and smartwatch processors has yet to pay off in a meaningful way, but now the company has another play: data centers. Yesterday, QCOM took the wraps off Qualcomm Cloud AI 100, a new processor aimed at data centers where artificial intelligence (AI) is increasingly important.
Qualcomm Announces Cloud AI 100
At an event on April 9, Qualcomm announced a new processor, targeting a new market for the company. The new Qualcomm Cloud AI 100 is aimed at data centers where AI is critical. The process known as “AI inferencing” is where already trained AI puts what it has learned to practical use. For example, AI inference is used to recognize images, to make recommendations to online shoppers or to translate speech into text.
Qualcomm’s approach is to use its expertise in designing the mobile processors used in smartphones, and apply this approach to AI. Smartphones require significant processing power in a very compact package, and it must be an extremely energy efficient design.
The Qualcomm Cloud AI 100 isn’t as powerful as the competition’s AI inference chips, but it has two key advantages carried over from that smartphone expertise: less heat generated, and much lower power consumption. One of the biggest operational expenses of a datacenter is powering and cooling all the processors inside. That operation is so energy intensive that data centers are predicted to consume 20% of global electricity production by 2025.
QCOM says its Cloud AI 100’s low energy requirements offers 10 times the power performance of competing chips. Expected to begin testing later this year with partners including Microsoft (NASDAQ:MSFT), they are scheduled to go into production in 2020. Qualcomm says that with its manufacturing capability, it will be able to ramp up output to meet demand very quickly.
Entering a Crowded Market With a Big Potential Payoff
The datacenter market is a big one, and within that broad category, the datacenter accelerator market — AI inference — was worth an estimated $2.84 billion in 2018. But it’s forecast to grow at a CAGR of 49.47% to hit $21.19 billion by 2023. That’s just short of Qualcomm’s $22.73 billion in total annual revenue for 2018.
The potential payoff for QCOM is big, but so is the competition. The Qualcomm Cloud AI 100 will be going up against established datacenter powerhouses Nvidia (NASDAQ:NVDA) and Intel (NASDAQ:INTC). In addition, Reuters points out that tech giants, including Amazon (NASDAQ:AMZN) and Alphabet’s (NASDAQ:GOOG, NASDAQ:GOOGL) Google are making their own AI inference chips.
QCOM says it is focusing on the smaller data centers that are proliferating globally as mobile apps and other connected services push for faster response times. These smaller “edge” data centers lack the cooling infrastructure needed to power endless racks of Nvidia or Intel chips, they tend to be more cost sensitive and most don’t need the same raw computational power a company like Google expects. The Qualcomm Cloud AI 100 with its high performance-per-watt is ideal for these edge data centers.
Qualcomm Cloud AI 100 Boosts the QCOM Outlook
QCOM’s focus on the smartphone market took Qualcomm stock to near record heights in 2014, but since then it has been up and down. Currently, QCOM is off those 2014 levels by about 30%, but it has been trending up and InvestorPlace’s Chris Tyler wrote it was a buy. That was a day before the Qualcomm Cloud AI 100 announcement. Add the company’s new push into the lucrative data center market, and QCOM’s future is definitely looking rosier.
As of this writing, Brad Moon did not hold a position in any of the aforementioned securities.
