Intel Eyes Fab 5G Future, Not Another Mobile Mess

It's probably not unfair to describe Intel's previous foray into mobile territory as an epic blunder. Despite marshaling its vast R&D resources, the semiconductor giant was unable to dislodge the UK's ARM as the microprocessor designer of choice in the 4G smartphone era. Branded Atom, its processors bombed. As quietly as it could, Intel shuttered the mobile chip shop and went in search of other business.

But now the company is back with big plans for 5G, and it's been making a lot of noise about the next-generation mobile technology several years before it's even due to hit the market. Is history about to repeat itself?

Intel Corp. (Nasdaq: INTC) obviously hopes not, and perhaps with good reason. For one thing, it has effectively given up trying to compete in the application processor space. "The 4G efforts were anchored on both the application processor and the modem, and we've staked our focus now on the modem side," says Sandra Rivera, the vice president of Intel's data center group and general manager of its network platforms division.

On the Intel Inside
Sandra Rivera, the vice president of Intel's data center group and general manager of its network platforms division, is betting on the potential of 5G.
Sandra Rivera, the vice president of Intel's data center group and general manager of its network platforms division, is betting on the potential of 5G.

The fruit so far is a 5G modem that Intel claims works across a range of sub-6GHz and very high frequency bands. In case anyone has forgotten, the sub-6GHz bands are ideal for wide-area and in-building coverage, while the much higher ranges will support the very zippiest services. Both seem bound to figure in operators' 5G plans. (See Intel Readies Multi-Use 5G Modems to Compete With Qualcomm.)

Announced in January, in the run-up to this year's Mobile World Congress, the full modem is due for trials in the second half of this year and could eventually make its way into an array of end-user devices. With 5G new radio specifications due to be frozen at the end of this year, and the first standardized services arriving in 2019, Rivera expects 5G devices to start shipping in meaningful quantities come 2020. (See 3GPP Approves Plans to Fast Track 5G NR.)

Besides narrowing its focus on the components side, Intel is trying to be more gregarious. "I think we were really going it alone during that [4G] time," says Rivera. WiMax, a "4G" network technology that Intel backed, was "superior" to LTE, she insists, but didn't catch on because of Intel's isolation. "We didn't bring the rest of the industry along and align with them," she says.

In the spirit of collaboration, Intel can today point to a handful of high-profile partnerships with other 5G stakeholders. That includes tie-ups with Sweden's Ericsson AB (Nasdaq: ERIC) and Finland's Nokia Corp. (NYSE: NOK), two of the world's three biggest equipment makers (China's Huawei Technologies Co. Ltd. being the other, of course). In mid-February, Nokia was said to have completed a "pre-standards" 5G connection using a data center based on Intel architecture and what the semiconductor giant describes as its "5G mobile trial platform."

It is perhaps the computing and architectural implications of 5G that could really make it a big deal for Intel. Operators like Germany's Deutsche Telekom AG (NYSE: DT) have talked about overhauling their networks and replacing big data centers with more distributed, "front-end" facilities. Doing so would minimize latency -- the delay that occurs in sending signals over data networks -- and could open up new service possibilities. But it would mean installing a lot more computing power at the very edge of the network. And that could mean a lot more business for Intel. "We are trying to drive more computing closer to the end point," says Rivera.

Virtualization and programmability look set to be key requirements in this overhaul as operators demand greater efficiency from their networks. But the more general transition from purpose-built "black boxes" to cloud technologies could play right into Intel's hands, allowing it to shift more of its off-the-shelf CPUs (central processing units). And as it targets emerging opportunities at the network "edge," Intel should be able to draw on its experiences in the core network area. "We've been there for many years and that part of our business is actually very large, relatively speaking, in terms of the network infrastructure group," says Rivera.

Next page: More ARM wrestling ahead

1 of 2
Next Page
KiptainLene 4/24/2017 | 6:31:19 AM
Fashion Best of Best really well said
mhui0 4/21/2017 | 9:40:46 PM
Re: I wish them well but They were not wrong. They were just early.
NorthwestTony 4/14/2017 | 7:01:17 PM
I wish them well but Intel is like the smartest kid in the class, who solves partial differential equations on the school bus for fun but doesn't have the social skills to play with the other kids.  When the college cheerleader asked him to take her to the prom, Otellini penciled it out and decided to pass on the iPhone.  In the 90s they bought DEC's ARM chip business for $700M, sold it 9 years later for $600M - then Atom got crushed by ARM chips.  Around the turn of the millenium they had the vision to start a global data center business from scratch, but lost their way and shut it down about 5 minutes before cloud computing became the next big thing.  It could have been Amazon Web Services!  And what in the end was Intel Media all about?  For a company that prides itself on Operational Excellence above all else, Intel has a horrible record executing its many good ideas that don't revolve around x86 fabrication.
danielcawrey 4/14/2017 | 12:34:00 PM
Atom The problem with Atom is that is required a specific form factor. Remember netbooks? They were the prime use case for Atom.

When compared to ARM's designs, that was a big problem. ARM chips didn't require a fan and were lower power when compared to Atom at the time. That's really why Intel couldn't get into the mobile market properly. 
Sign In