Intel and Altera Are All About Energy

Brian Hicks

Posted June 1, 2015

If you’re paying attention to the trends in enterprise computing, today is kind of a milestone day.

Leading North American chipmaker Intel (NASDAQ: INTC) is finally coming to an agreement to acquire Altera Corporation (NASDAQ: ALTR), a senior manufacturer of reconfigurable digital circuits known as programmable logic devices (PLDs).

Murmurs of the two companies merging first began last year, and the rumors kicked into high gear just a few months ago when Altera rejected a bid from Intel.

Trading of both INTC and ALTR was halted on Monday morning following the announcement that the $15 to $17 billion merger would break out to Intel paying approximately $54 per share of Altera. This value would be approximately 55% higher than Altera’s share value two months ago.

While the deal has not been sealed as of the moment I’m writing this, the merger is an inevitability shaped by a consolidation trend in server-side processing.

This is a tectonic shift that must be acknowledged for any investment in chips and processors.

A marriage in the name of energy

Back in 2014, Intel announced it was integrating its Xeon processor with an as-of-yet unnamed FPGA (field-programmable gate array) into a single package. Intel representatives said users of this package could expect their application performance to increase 10-fold with only a minor increase in power consumption.

The science to back up these claims came from Microsoft, who developed a data center system that married FPGAs and processors to speed up the performance of its search engine Bing.

The big selling point of this marriage is that it cuts down on energy requirements. Data centers currently consume as much as 10% of the world’s total electricity. That’s kind of a staggering thought, really, when considering that the entire transportation sector (both commercial and personal) consumes 20%.

And data center energy consumption is expected to continue its huge growth. We’re not getting the dramatic leaps in processing power that we used to, and the focus is shifting to decreasing energy demands.

In 2011, Facebook debuted its own designs for server hardware after it spent two years redesigning its Oregon data center. This initiative, known as the Open Compute Project, strove to design the most efficient servers, storage systems, and data centers in an open environment so the cost of infrastructure computing could drop and its environmental impact would improve.

Just two months ago, Apple, Cisco, and Juniper Networks joined the project.

At around the same time in 2011, big tech companies began toying with ARM-based processors in servers.

ARM processors utilize a reduced instruction set (RISC) that allows for higher-efficiency computing. It’s the architecture that dominates the battery-driven world of smartphones and tablets where energy efficiency is a top priority.

Computers and traditional servers, meanwhile, use the x86 architecture, which utilizes a complex instruction set (CISC). Companies thought they might be able to cut down on energy demands if they built servers using RISC-based processors.

Since Intel’s strength is in x86-based processors, experiments in ARM-based servers looked like an attack on Intel’s incumbency.

This is still a developing field. Four years ago, Hewlett-Packard unveiled “Project Moonshot,” which was a promising cluster of low-energy ARM-based servers. Unfortunately, it wasn’t much of a hammer when it hit the market last year.

Early in 2015, Chinese computer giant Lenovo announced it was building a prototype ARM-based server in the interest of making more efficient data centers and improving performance per watt/performance per dollar overall.

In its recent quarterly earnings call, Applied Micro (NASDAQ: AMCC) proudly trumpeted the fact that eBay’s payment subsidiary PayPal uses its ARM-based X-Gene servers in its data center.

Applied Micro’s CEO Paramesh Gopi called the server business a “single architecture monopoly.”

If Intel’s new FPGA/processor architecture can massively improve the energy demands of x86-based servers without simultaneously increasing the difficulty of programmability, then the monopoly might continue unchanged.

Until then, the quest for a more efficient data center will continue to leave the door open to smaller competitors.

Good Investing,

  Tim Conneally Sig

Tim Conneally

follow basic @TimConneally on Twitter

For the last seven years, Tim Conneally has covered the world of mobile and wireless technology, enterprise software, network hardware, and next generation consumer technology. Tim has previously written for long-running software news outlet Betanews and for financial media powerhouse Forbes.

Angel Publishing Investor Club Discord - Chat Now

Brian Hicks Premium

Introductory