Qualcomm (NASDAQ:QCOM) launched its first line of Centriq data center chips last November. The future looked promising at first -- Microsoft started testing the Centriq chips in its data centers, and initial benchmarks indicated that the chips could be viable alternatives to Intel's Xeons.
But in early May, multiple reports indicated that Qualcomm could abandon the data center market. Later that month, Qualcomm's data center technologies chief Anand Chandrasekher, a former Intel (NASDAQ:INTC) executive, resigned. Qualcomm downsized its data center unit in June but denied that it was exiting the market.
However, several rounds of layoffs, including one in early December, reduced the size of Qualcomm's data center technologies group from roughly 1,000 employees to about 50. The company claims that its data center group isn't dead yet since its chips will still be used in its HXT server joint venture in China, but the writing is clearly on the wall. Let's examine the four factors that killed this once-promising business.
1. Trying to dent a near monopoly
Intel controls over 95% of the world's data center CPU market with its Xeon chips, which have been the industry standard for over two decades. This has made it tough for smaller chipmakers to gain any meaningful traction in the market.
Qualcomm thought that it could leverage its leading position in ARM-based mobile chips to expand into the data center market. However, Arm Holding's chip designs are generally associated with more power-efficiency though lower horsepower output than Intel's x86 designs. The Centriq marked a promising leap forward for ARM-based servers, but not enough data center operators were willing to make the switch.
2. AMD's better alternative
Advanced Micro Devices (NASDAQ:AMD), Intel's only meaningful rival in the x86 market, reentered the data center market last year with its Epyc CPUs. Most benchmarks showed that the Epyc CPUs offered performance comparable to that of Intel's Xeons at much lower prices -- an appealing proposition for some data center operators, who wanted cheaper x86 chips instead of ARM-based ones.
Intel's market share in data centers reportedly fell from 98.6% to 96.6% after AMD entered the market, according to IDC. Earlier this year, Intel's then-CEO Brian Krzanich admitted that the company's goal was to merely prevent AMD from claiming a "15% to 20%" share of the data center market instead of shutting it out. This indicates that neither Intel nor AMD (which also sells ARM-based server chips) considers Qualcomm a meaningful rival in the data center market.
3. Broadcom's hostile bid for Qualcomm
Broadcom (NASDAQ:AVGO) made a hostile bid for Qualcomm in late 2017, which was ultimately blocked by an eleventh-hour intervention from the Trump administration in March of this year. But before that happened, Qualcomm tried to rebuff Broadcom's offer and appease its investors with drastic cost-cutting plans to boost its earnings.
Qualcomm cut its spending by $1 billion, and the data center unit bore the brunt of those cuts. This move crippled the fledgling unit just as AMD's Epyc chips gained market share against Intel's Xeons.
4. Failed buyout attempts
Prior to his departure, Chandrasekher tried to convince a group of investors, including Arm Holding's parent company SoftBank Group (NASDAQOTH:SFTBY) and Singapore's Temasek Holdings, to buy Qualcomm's data center unit. Those talks collapsed after Qualcomm CFO George Davis set a tight deadline for a deal.
Qualcomm's former CEO and chairman Paul Jacobs then tried to buy the data center unit after he was ousted from the company in March. However, Qualcomm asked Jacobs to halt his efforts to also take the entire chipmaker private as part of the deal -- which he refused to do.
Former Intel executive Renee James also considered buying the unit and integrating it into her data center chip start-up, Ampere Computing, but walked away due to the design-sharing conditions of Qualcomm's data center joint venture in China. After all those failed buyout attempts, it looks as if the data center unit is destined to wither away and become a footnote in Qualcomm's history.
What does this mean for Qualcomm?
Qualcomm hoped that expanding into data center chips would reduce its dependence on the mobile chip market, where it faces a saturated environment filled with tougher competition. Defiant original equipment manufacturers, like Apple, are also dumping its mobile chips and refusing to pay licensing fees for its wireless technologies.
The loss of the data center unit, along with its failed takeover of automotive chipmaker NXP Semiconductors, dramatically reduces Qualcomm's long-term growth opportunities. That's a shame, because its data center unit might have survived with the proper funding or a sale to the right suitor.
Teresa Kersten, an employee of LinkedIn, a Microsoft subsidiary, is a member of The Motley Fool's board of directors. Leo Sun owns shares of Apple. The Motley Fool owns shares of and recommends Apple. The Motley Fool owns shares of Microsoft and Qualcomm and has the following options: long January 2020 $150 calls on Apple and short January 2020 $155 calls on Apple. The Motley Fool recommends Broadcom Ltd and NXP Semiconductors. The Motley Fool has a disclosure policy.