x
Comms chips

Intel CTO: The Future Is Here!

ATLANTA -- Supercomm -- If you looking for some refreshing optimism amid the gloom and doom of the current telecom economy, it's probably a good idea to catch up with Eric Mentzer, vice president and CTO of Intel Corp.'s (Nasdaq: INTC) Communications Group.

Mentzer, in charge of piloting Intel's technical vision for the future of communications chips and components, says the optical revolution has just gotten started -- and that it's likely to power technology and bandwidth growth through the next ten years. At the heart of it will be a high-powered fusion of silicon and optical technologies.

"We've gone through ten years of change in two years," says Mentzer on the show floor here at Supercomm 2002. "It's just a sea change."

Mentzer quickly sketches some diagrams, obliquely mentioning something he calls "Eric's Law," in which processors must double their operations power at the same time that they double their speed. He then sketches out the "modular network," in which wide-area networks connect to a Web of wireless networks, home networks, and enterprise networks.

What's the big picture? Through the integration of silicon and optical components, high-speed communications technology is quickly being commoditized and working its way down the food chain, from the largest carrier networks to home networking. The net result?

"In ten years optics will come on every desktop chipset," Mentzer says.

Better yet, though, Mentzer believes the pain that telecommunications carriers are going through now has a purpose: It's forcing them to migrate from highly specialized, proprietary networks to those based on open data standards, which provided the explosion on corporate networks. And he believes this transition in carrier technology will eventually yield cheaper, more efficient networks that will return the carriers to profitability. The emerging standards include those for Gigabit Ethernet.

"Ethernet in the carrier network will bring the benefits of reducing operating expenses. It's these kinds of open standards that will bring us out of the recession. The world has changed now; there's no going back."

What does all this mean for Intel? According to Mentzer, it means that Intel's Comm Group is focusing on a set of highly integrated components based on 90nm technology. He calls this a "unifying technology" that will produce highly integrated modules that, for example, could combine wireless communications, Ethernet, and high-speed optics capabilities in a single chip.

Though Mentzer didn't mention specific products, he did reveal that one of Intel's upcoming announcements will include a 90nm, 2GHz network processor that processes 60 million packets per second.

That's just the baby step toward the ten-year plan, he says. "We'll have processors running at 10 GHz in ten years."

If Mentzer is correct, the doldrums of 2002 may soon be yesterday's news.

— R. Scott Raynovich, US Editor, Light Reading
http://www.lightreading.com For more information on Supercomm 2002, please visit: Supercomm Special
Page 1 / 2   >   >>
edgecore 12/4/2012 | 10:18:27 PM
re: Intel CTO: The Future Is Here! -------------
Though Mentzer didn't mention specific products, he did reveal that one of Intel's upcoming announcements will include a 90nm, 2GHz network processor that processes 60 million packets per second.
-------------


Nice to see someone predicting 10 years out, we can barely tell what the hell is going on 10 DAYS out.

Anyways, who wants or needs 2/5/10 Ghz Intel, unless you want to BBQ a steak on your router.

I predict that in 10 years we will all have OC768 PON's to our homes delivering 3D scratch and sniff holograms of Britanny Spears...can I have Chambers's keynote slot tomorrow morning to discuss further?

EC
hyperunner 12/4/2012 | 10:18:25 PM
re: Intel CTO: The Future Is Here!
"If Mentzer is correct, the doldrums of 2002 may soon be yesterday's news."

Sorry, Scott, but I don't see it. Our current doldrums are caused by the fact that we (the industry) have built out a huge network infrastructure over the past decade without having any idea who will pay for it.

So the way to get out of the doldrums is to allow carriers and SPs to charge a "living margin" for the services they offer. If market forces were allowed to operate (ie. without regulatory meddling), then we should be seeing prices for services going up.

But on the contrary, we're seeing voice pricing continue to drop, even though most carriers are now at the point where they're losing money on their former "cash cow".

I don't see a data service coming up that can replace the massive revenues of voice (the scratch 'n sniff Britney app would certainly be a "killer", but only in terms of what my wife would do to me if she caught me using it :-)

But seriously, go find the most optimistic worldwide forecasts for "emerging data services" (eg. IP VPNs, EPL, TLS, etc.) and add up their total revenues. I bet they don't even beat the voice revenues from somewhere like Luxembourg for last year.

Our problem as a carrier is how we're going to turn our voice revenues back into profitable numbers. Any ideas? Please don't say "VoIP", because "pay to save" is not too popular with our CFO. The best strategy we came up with so far is "invite the regulator to lunch in the staff cafeteria and make 'em eat those scary white German sausages until they agree to get off our case".

hR
opt_dhogan 12/4/2012 | 10:18:25 PM
re: Intel CTO: The Future Is Here! *********************************************
Sorry, Scott, but I don't see it. Our current doldrums are caused by the fact that we (the industry) have built out a huge network infrastructure over the past decade without having any idea who will pay for it.
*********************************************

I hate to beat a dead horse, but...

Let's stop thinking about finding ways to pump more data into my home or reasons to do it. Let's find new ways to charge me for data (voice, video). I'm not a very creative person so other's ideas will be better than mine. How about video phone or streaming VoD to start?

Hyperunner says the industry didn't have any idea who would pay for the network. Let's start finding that money, not looking for new dreams of high bandwidth technology or applications.

-d
hyperunner 12/4/2012 | 10:18:23 PM
re: Intel CTO: The Future Is Here! Hi d,

Believe me, I'm with you on looking for new ways to visualise this problem, but it's hard (I guess if it was easy we would have solved this by now).

You've chosen two interesting examples; Videophone and VoD.

There's a great Dilbert cartoon (I have it on hard copy but maybe I'll scan it for you..). Dilber buys a videophone, the first guy to do so. The last pane of the cartoon sees him sitting at the videophone, waiting for somebody else to buy one so he can call. OK, it's not that desperate, in fact now I'm in Europe I can tell you that video-over-IP is hot stuff here - lots of old PicturTel ISDN-based VC consoles are being thrown out and replaced by video-over-IP. The cost model is very, very compelling, but every single packet is passing over VPNs (typically IPSec, not MPLS). People can't do inter-company IP-VC because the firewalls on their site, or sometimes in the carrier network just don't pass the packets. Even if they get those issues sorted out, Network Address Translation prevents true, peer to peer conversations happening (we need a Napster for IP Videoconferencing). And even if that happens, IP-VC puts a big load on the network - it's real-time, after all, and there's no such thing as IP QoS (don't let those Cisco and Juniper guys fool you :-)

For VoD you don't need stringent QoS, because it's not real time. You can put huge buffers in the network to take care of congestion, and you can do clever things with local caches, and if I combine that with emerging DiVX encoding we think that 1Mbps will get you a VHS-quality VoD stream real soon now. But who's going to pay for VoD when you can hire a video from the local gas (sorry petrol) station for $1.50? I have digital satellite to my home now with Sky TV, and if I really want a "movie on demand" I can pay over a phone line, and get a "Dolby Digital" movie, or a sports event within a few minutes of any start time. And I don't need to wait for the incumbent Telco to lay on an ADSL line to my (rural) home.

I'm happy to be challenged on the VoD business case, but it looks bleak to me for Europe. European MSOs are hurting bad on most of their businesses. NTL in the UK (who were close to Chapter 11 recently) are losing money on their video and Internet service. Voice is the only service where they're making money (not for long, believe me). I know N.America is different. Been away too long to be "current".

hR.
sauron5 12/4/2012 | 10:18:20 PM
re: Intel CTO: The Future Is Here! Well, one thing the ICTO got right. Enterprise
can't support growth in Optical Networking. That
market is high margin, small, and currently has
overcapacity for 5 years (according to some
unimaginative analysts. I could use all that bandwidth in 5 days) OTOH, the average home
user isn't going to pay more than $79 dollars/mth
for broadband connections while still paying
$30 for PSTN and $30 for Cable TV and $30 for a
Wireless Phone.

Sauron
silly_valley_boy 12/4/2012 | 10:18:18 PM
re: Intel CTO: The Future Is Here! Good analysis making the rounds...
-----------------------------------

Intel has no choice but to exit the communication IC business. A simple analysis of their annual report data shows that their stock could be 2x where it is today, had they simply shut down the two divisions that are bleeding badly.

In 2000, Intel's revenues were $33.7B and they posted an earnings per share of $1.51. Intel's stock started the year at around $30 and exited the year at around $60 (split adjusted). In 2001, Intel's revenues dropped to $26.5B and they posted earnings per share of $0.19. The stock had dropped to $30 by April 2002. Like most companies, Intel has experienced the pinch of the current recession. But unlike most companies, Intel has yet to close the divisions that are bleeding the worst. Let's have a look at the 2001 Annual Report numbers to see what they are likely to do.

Intel is broken into three main divisions:
- The Intel Architecture Group (IAG) makes Pentiums and chipsets for PCs as well as a few consumer products.
- The Intel Communications Group (ICG) makes network processors, embedded controllers, and MAC/PHY/optical components.
- The Wireless Communications and Computing Group (WCCG) makes wireless chips and Flash memory (the bulk of the revenue here is Flash).

Here's the revenue/profit/loss breakdown from each division:

Group Revenue Profit/Loss % of Total % of Loss
Revenue
--------------------------------------------------------------------------------------------------------------------------------------
IAD (Pentiums) $22,195M $6,300M 82% n/a
ICG (Comm ICs) $ 2,570M ($ 735M) 9.5% 74%
WCCG (Flash) $ 2,294M ($ 256M) 8.5% 26%

[All numbers from the 2001 Intel Annual Report]

The combination of the two communications components groups lost a staggering $991M for Intel in 2001. With 6.7B shares outstanding, this represented a loss of $0.15/share. In other words, had Intel simply shut down these divisions in 2001, they would have earned $0.34/share vs. the $0.19 they posted (it would actually have been less than this for write-offs, etc. but you get the point). This could have meant double their current stock price.

If I were the CFO over at Intel, I'd be looking long and hard at these bleeding divisions, especially those that don't promise to be anywhere near as big as the Pentium, even in good years. As CEO, how do I justify to shareholders a $1B loss in a market that will never offer the volumes of the PC business? I can't.

One obvious target at Intel is the NPU business. The Intel Communications Group has over 1,000 employees on NPUs alone. This equates to a conservative cash burn of $200M per year ( a significant amount of the $735M lost in 2001). Burning that much cash *must* equate to a big market potential, right? Wrong. Analysts put the NPU market at an optimistic *maximum* of ~$350M/yr in 2003. If Intel got 100% market share (virtually impossible in this market) they'd still contribute only around 1.5% of Intel's overall revenue and net just $150M. They'd have to get dominating market share just to break even.

And Intel has never had dominating market share in anything other than PC CPUs.

With a string of bad acquisitions, followed by staggering losses, it is only a matter of time before they realize where the money is. And where it isn't.
dbostan 12/4/2012 | 10:18:18 PM
re: Intel CTO: The Future Is Here! How about they will realize where money WILL BE in the future?
I believe, that Intel should actually expand their investments in the two mentioned divisions, given the technological migration from desktops to networked (wireline and wireless) devices.
strands555 12/4/2012 | 10:18:17 PM
re: Intel CTO: The Future Is Here! silly_boy aka AMD_boy:

This is the 3rd time you've posted that same message in the past 6 weeks. Agenda maybe???

Using the same short-sighted arguments, in the midst of the most severe recession ever in the sector, every single comm and networking company should close down every new growth division, laying off tens of thousands more people, then sit on their collective thumbs praying that their legacy lines carry them through AND become the future. Ridiculous.

If you don't believe in the future growth of the industry, you should change industries. If you DO believe that future growth will return, but don't invest in that future growth during the lows of the cycle, your actions don't match your words. You will be left behind when the growth does return because all you'll have is legacy product lines. It is really that simple. If you don't have a full pipeline of new products and business lines, you'll be toasted by those that do, when the growth returns.

I commend all companies that are betting big on future growth right now. They will be the reason for the momentum coming off the lows. If all companies were in "please wall street this quarter" mode we would never come off the lows because there is no fuel in the pipeline to give any reason to believe there will be growth.
sauron5 12/4/2012 | 10:18:16 PM
re: Intel CTO: The Future Is Here! Hey, if it's a choice between pay for employees
or pay for the R&D for the next generation, I think
I am going to have to go with pay the employees.
That's why startups left and right are closing
their doors. There is more of this to come.
Capitulation will not solve the problem. The
$675 Billion invested by CV is gone and not coming
back. So, where is the money for the future?

Sauron
fusionboy 12/4/2012 | 10:18:07 PM
re: Intel CTO: The Future Is Here! opt_dhogan said:

"Let's stop thinking about finding ways to pump more data into my home or reasons to do it. Let's find new ways to charge me for data (voice, video). I'm not a very creative person so other's ideas will be better than mine. How about video phone or streaming VoD to start?

Hyperunner says the industry didn't have any idea who would pay for the network. Let's start finding that money, not looking for new dreams of high bandwidth technology or applications."


The reason we're all looking for brand new big fancy applications is this has historically, been the only way to get people to pay substantially more for any service upgrade. Look at the sqawk recently over the upped long distance rates - and this was only an increase of 10-25%. Neither of the applications you mentioned are going to get people to pay substantially more money and here's why:

- VOD - when you can buy the movie on DVD for under $20(for a topnotch new release), or rent as many as you want a month for $20 from Netflix, your going to have to cut your price pretty low to get takers. Unless your trying to go for the sports market - but how big is the market that isn't served by Satelite? I can get every NFL game with DirectTV - sure some people want college games - but how many of them are there?

-Videophone - is dead. Will be dead. Sure it's a nice gadget, but the cost won't drop low enough for a long time, except for PC usuage. And we've seen how much you can charge people for home bandwidth.

In short their aren't many new ways to charge people for what they dont' already use - and there aren't many new things that they want, that you can charge them for. Of course this will change, but not soon.
Page 1 / 2   >   >>
HOME
Sign In
SEARCH
CLOSE
MORE
CLOSE