The news last week of Qualcomm's renewed interest into the buzzing ARM-based server market got me thinking: should Qualcomm, the Intel of the ARM ecosystem, buy AMD, the only other significant x86 player in the market - and an ARM licensee?
AMD's fortunes have fluctuated as wildly as a seismograph over the past decade because it failed to react promptly to changes in the market and to Intel's relentless challenge in the x86 market.
Since the beginning of the millennium, its shares have lost more than 90% of their values and the only reason why Intel didn't squeeze it out of the market completely is because of potential antitrust/monopoly concerns in the US.
Moving away from the traditional market
Furthermore, only die-hard supporters would bet on an improbable AMD revival in the x86 CPU market where Intel outrageously dominates benchmarks and sales.
The days of the Athlon and the K8 architecture being compelling alternatives to Intel's products are long gone. AMD simply doesn't have the resources (financial or human) to compete with Intel at this end of the spectrum.
The Sunnyvale-based company however made three moves that might have secured its long-term future. Firstly, it got rid of its chip fabs, then it purchased Canadian graphics chip maker ATI and thirdly, it took an ARM license. All three combined prepare for a future beyond the traditional x86 processor.
As for the "why", if Qualcomm wants to succeed in the server/enterprise market, it needs time; Samsung, Calxeda and Nvidia have pulled out of the race, leaving AMD as the only serious contender ahead of Broadcom, Cavium, Marvell and Applied Micro.
It's all about the server
AMD built its server reputation from ground up with the Opteron range, launched in 2003, and managed to secure significant wins. Sledgehammer opened a lot of doors; IBM, HP, Dell amongst others became long-term partners, the sort of client that Qualcomm would like to have on its customer list.
Qualcomm can emulate AMD but doing so will take a lot of time. Other than the usual FUD (fear, uncertainty, doubt) factor, there's the fact that Qualcomm is an unknown quantity for the Big Iron manufacturers plus many would need to be convinced about an architecture shift (x86 to ARM).
Interestingly, AMD was the last one to have convinced manufacturers to do so when it introduced x86-64 back in 2003 and it is doing a good job with its APUs.
A $3 billion shortcut?
The company is currently valued at about $2.9 billion (£1.71 billion, AU$3 billion), a fraction of its much bigger competitor, Intel or even Qualcomm. The latter has always been keen on acquisitions; it bought AMD's handset division back in 2009 and Atheros, its biggest purchase back in 2001, was worth more than AMD.
Latest financials show that Qualcomm has just under $15 billion (about £9 billion, AU$ 15 billion) cash and short investment interests so money is not an issue.
Buying AMD would give it not only access to an impressive portfolio of customers and some of the best engineers in the world but also a head start both in servers and in graphics.
Qualcomm has been mooted as a potential suitor the last time AMD was apparently put up for sale back in 2011, back then though, the company was worth three times more and Qualcomm had much less liquidity.