Can Intel's 'Tick-Tock' Road Map Survive The Recession?

Microprocessor giant Intel prides itself on pushing out cutting-edge technology at a lightning-fast pace. But the current economy calls into question whether the vendor's strategy still makes sense.

After falling behind smaller rival Advanced Micro Devices on multicore technology several years ago, Intel unleashed a new plan to set the pace of innovation by tightening product development cycles and hitting yearly road map targets with metronomic precision.

They call the strategy "tick-tock." And for the past several product cycles, it's worked a treat, delivering dominant market share and attendant profits to Intel even as AMD and other chip makers struggled to keep up. Spearheaded by CEO Paul Otellini, Intel's leadership demanded flawless product road map execution, and the company's engineers delivered.

But can Intel keep it up?

Few would doubt that the Santa Clara, Calif.-based chip maker has the organization and the brainpower to nail whatever road map targets it sets. But execution isn't the issue. The question is whether Intel has the kind of mutant marketing superpowers that would be required to continue shipping the most expensive new processors and platforms in a climate of severe cost-cutting, project postponements and all-around economic uncertainty.

The chip giant has deep pockets and could theoretically bull its way to the recovery without changing a thing. But some Intel channel partners, most of whom have considerably less wiggle room, say they're concerned that the company's normative "tear-down, start-from-scratch" product cycle isn't appropriate for current market conditions -- or the current needs of the system integrator channel.

And is it really in Intel's interests to keep driving in the fast lane when everybody else is pulling off the highway? Can "tick-tock" survive the worst economy since the Great Depression?

Bad Numbers, Worse Guidance

The semiconductor industry has been hit hard and fast by the global economic downturn. Intel is no exception -- the chip maker saw its fourth-quarter revenues drop 19 percent compared with the previous quarter, the first time in years Intel did worse in Q4 than in Q3. The profits picture was even more frightening: Intel's net income in the last quarter of 2008 declined a whopping 90 percent vs. the fourth quarter of 2007.

And that's not the end of it. In fact, it's going to get worse, promised a less-than-jaunty Otellini during Intel's recent fourth-quarter earnings call with analysts. Intel wouldn't provide formal earnings guidance for the current quarter, but Otellini said that "for internal purposes, the company is currently planning for revenue in the vicinity of $7 billion" -- which would be a 28 percent decline from the $9.7 billion Intel hauled in for the first quarter of 2008.

How bad will it get? Bloomberg last week cited an internal Intel memo in which Otellini reportedly told employees that the company's prospects for finishing the first quarter in the black were "too close to call."

Intel has turned a profit for 87 straight quarters, even surviving the dot-com implosion in the early part of this decade without going into the red. Don't think this recession is for real? Just imagine that remarkable streak of profitability ending.

"We are not going to wake up in six months with everything rosy again," Otellini wrote in the memo.

Against this bleak backdrop, it's understandable that Intel recently pulled out one of the most reliable weapons in its inventory-thinning arsenal -- slashing processor prices. The company cut prices by as much as 48 percent for some 20 older chips in its lineup.

It's the old quandary: The market isn't buying, inventory is sitting around costing companies money and it's become crucial to get parts out the door by any means possible. "Inventory is king," said Otellini during the earnings call.

That's a message that hits home with companies downstream from Intel. Everyone is seeing demand for new hardware dry up, from the PC giants like Hewlett-Packard and Dell to the smaller whitebox players in the channel that make up some 40 percent of Intel's global business. The computer makers, struggling to clear inventory, have passed along their pain to the component manufacturers like a punch to the gut, hence the flurry of price cuts from the likes of Intel and Sunnyvale, Calif.-based AMD. Slashing prices for older products does have a lot of appeal for channel partners, as much needed leverage to help pry open locked-down consumer and business pocketbooks, said longtime Intel partner Joe Toste, vice president of marketing at Minneapolis-based whitebox builder Equus Computer Systems.

But such efforts to jump-start the market, coupled with an apparent lack of demand for Intel's higher-priced new chips and platforms, is also going to mean drastically falling margins -- for both the chip giant and its channel, Toste said.

At least it's the sort of pain Intel was up-front about on the recent earnings call. Intel executives warned of gross margins dropping from 53 percent in Q4 to "the low 40s" in the current period.

"Regardless of how Intel plans to roll out new technology, they have no choice but to go through these price bands at some point," Toste said. "I don't see that fundamentally changing for Intel or AMD. The difficulty is, when you have a strong economy, they can make more margin by keeping a part at a higher price for longer. You can keep a price point going for four, five months. Now they have to drop prices after two or three months."

Intel also recently announced the planned shuttering of five of its older manufacturing facilities by the end of 2009 -- actions that will affect between 5,000 to 6,000 employees, or about 7 percent of its current workforce.

Such moves would have big repercussions in ordinary times, but again, these may not be ordinary times. Will price cuts and fab shutdowns be enough for Intel to find some rosy morning to wake up to, sooner rather than later?

Next: The Tick And The Tock Of It

"Tick-tock" has been a favorite talking point for Intel executives, at the company's own Intel Developer Forums, partner events and on earnings calls, though the expression was noticeably absent last week when the company disclosed financial results. But "tick-tock," for all its marketing heft, is only vaguely understood by many in the industry, according to analyst Jon Peddie, president of Jon Peddie Research.

The first part, "tick," refers to various improvements to Intel's silicon fabrication process -- most famously, the roughly bi-annual doubling of transistor counts on integrated circuits, in line with Intel co-founder Gordon Moore's eponymous "law" that was first spelled out in a 1965 magazine article. Intel's unveiling of 45-nanometer processors in late 2007 and planned release of 32nm chips by the end of this year are examples of "tick" in action.

"Tock" is a more recently quantified process. For Intel, it describes major architectural upgrades made to the company's most prominent client and server processors, and the hardware platforms that support them. The chip maker's new Nehalem-class Core i7 chips, featuring a major reconfiguration of the memory controller, are the "tock" end of the equation.

Given Intel's publicity push for "tick-tock," many people mistakenly regard it as a set of goals Intel seeks to accomplish, Peddie said. In reality, "tick-tock" is more of a description of the only really viable manufacturing process for the semiconductor industry.

"Intel has been doing this 'tick-tock' thing, as has any other semiconductor company, for quite some time. It's just that it was Intel that put a name on it and identified it as a process," Peddie said.

"Intel has two types of investment it has to make. The fab is first. Second, there are the innovative designs to exploit manufacturing capabilities. Companies that broke this rule are no longer in business," he said.

Jim McGregor, an analyst with In-Stat, hammers that point home. Asked whether the "tick" part of Intel's road map, with its direct tie-in to Moore's Law, is "in Intel's DNA," McGregor dismissed the notion as the wrong way of looking at it: "Process shrinks aren't just in Intel's DNA, it's their business model. They cut their own costs when they do them."

Yet the questions aren't so much about the necessity of a "tick-tock"-style improvement process -- Peddie points to the flailing U.S. auto industry as an object lesson in not keeping up -- it's the pace and scope of Intel's product development cycle that seems increasingly difficult to maintain. Those Core i7s are remarkable products, yet these days Intel seems to be having a lot more success selling its ultra-low power Atom chips, which may never get the Nehalem "tock" upgrade.

Intel's reputation is one of relentless, machine-like precision. But the bottom line is this: Intel executives are beginning to send signals that they're prepared to be a lot more flexible with their plans in the face of a sustained economic crisis.

Next: Turning Back The Clock

With its success and growth, Intel has already somewhat diluted the message of "tick-tock." The Atom family, a new lineup of embedded processors, the upcoming x86-based "Larrabee" graphics processors -- none of these products slides neatly into the "tick-tock" development cycle, because they won't be transitioned to the Nehalem architecture.

On the other hand, even older brands like the Pentium have been transitioned to the most recent "tick" node, the 45nm die shrink, in recent months. That translates to much better performance from a stable, familiar product and it raises an interesting question for Intel -- why not maintain the "tick" cycle's pace while reducing resources spent driving the "tock" cycle?

Let's assume that going forward, Intel finds itself relying more and more on older and/or less pricey product lines like Pentium and Atom. It seems like it would make a lot more sense to keep pushing the process shrinks aggressively while de-emphasizing the cutting-edge microarchitecture upgrades. That's because the "tock"-cycle products generally require costly platform upgrades -- as is evident with Nehalem -- and the market just isn't ready to bear such costs.

At any rate, Peddie said it was never his understanding that "tocks" like Nehalem were supposed to affect all or most of the products in Intel's huge processor stable, but he said the chip giant has always been rather vague on that subject. What's clear is that in the past, Intel has "waterfalled" such microarchitecture upgrades over time from its top server and desktop parts down into most of its cheaper processor families.

But that's not going to happen with Nehalem.

"We're designing now for price and not waterfalling every new technology down through all the product families," said Steve Dallman, general manager of Intel's worldwide Reseller Channel Organization (RCO).

"Intel isn't robotic. Our plans currently are for more than one 32-nanometer fab. We're starting to pay for that 32nm process node now; that's why Paul said as much [on Intel's fourth-quarter earnings call]," he said.

With the Nehalem target reached, the transition to that process node is next, and 32nm products, code-named Westmere, are expected to first appear by the end of this year. But Intel CFO Stacy Smith, speaking on the earnings call, said Intel would "absolutely modulate the ramp rate of the process depending on demand."

Dallman wouldn't speculate about what that would look like precisely but did address the possibility of a very long recession.

"My gut tells me that if the world stagnates, do you need two 32nm fabs? I don't know. I don't see us not coming out with all those products," he said. "There are a lot of ways you can modulate a fab."

Of course, Intel could also choose to accelerate older processor lines to 32nm if the situation warrants, Smith said. Put that statement together with Intel's stated plans to shut down five older fabs this year, and a way forward starts to emerge. How about Core 2 or even Pentium at 32nm for cost-conscious system integrators looking for every penny they can save?

"I think most of the products we have right now on the Core 2 architecture were planned to be shipping in significant volume for years from now," Dallman said. "That's on 45nm now, which is going to go on for a long time. I don't think it's going to go to 32nm, but it could if they wanted to. Hey, Pentium is on the 45nm process now."

Whether or not Intel tweaks its road map here and there, Dallman is adamant about one thing: "We're going to execute on time."

"We're going to launch the Nehalem-based server chips in Q1. We're going time-to-market in the channel and we're not going to back down on that. New products lead you out of the recession," he said.

"We as a corporation know that you need to continue to invest during the downturn. I've been here 30 years. Every CEO has said those same words. This CEO is not changing those words," Dallman said.

But there are other areas besides just the CPU schedule where observers speculate Intel may have to adapt strategies to fit the market. These include the planned transition to DDR3-only products, which Intel is rumored to be pushing back by a few months.

Meanwhile, the pending retirement of former CEO and current Board Chairman Craig Barrett has sparked some speculation as well.

Intel, the theory goes, is planning to cut back on its manufacturing investments and R&D expenditures. Barrett, "a manufacturing guy," is rumored to have chosen retirement "rather than preside over a de-emphasis on Intel's manufacturing investment," reports ElectronicsWeekly.

In short, there's a lot of gossip out there. But where does all this leave the rest of us?

Next: Channel Conversations

Intel's ability to deliver new technology at the place and time it promised has been warmly regarded by its partner channel over the past few years of relative market stability. But as those same system integrators find selling costly new product more difficult, they're discovering a renewed appreciation for Intel's back catalog.

"The Core i7 is a phenomenal technology, so it should have some new chipsets around it. But in this climate, mature technology is more attractive. Building a new platform is hard to do," said Bill Paschick, president of Rain Recording, a Ringwood, N.J.-based builder of high-performance desktop and mobile audio workstations.

"As a system builder, as a manufacturer, even as a consumer, if you can help me make some cash by making use of things I have laying around, the more I love you. Everybody's having inventory issues, and the herd is thinning everywhere you look," Paschick said.

Paschick said he appreciated the recent price cuts to older processors like the Core 2 Quad parts, but also was seeing a different attitude from Intel with regard to its new Nehalem-class products.

"Intel is kind of infamous for tear-down, start-from-scratch updates. So you get the new stuff and it's, 'Throw out all the old stuff.' But they didn't totally step all over last year's technology with the Core i7," Paschick said. "The 'how many can I put you down for?' attitude is not going to work anymore."

Intel's high degree of visibility in the channel is ordinarily a huge boon to the company when it comes to moving new product, but these aren't ordinary times, said Equus Computer Systems' Toste.

"Intel likes the channel because they rely on system builders to ramp new technology like the Core i7 faster than the Tier 1 [OEMs]. Then it goes mainstream and the Tier 1s beat us up. Which is fine, because the channel in the past has been able to do its part at higher prices, but now, not so much," he said.

The "big gap" Equus enjoys in selling new Intel parts at a higher margin is narrowing fast in the current market. It's a snowball effect, said Toste, echoing Otellini's earnings call pronouncement, because "inventory is still king."

To that end, Paschick is happy to see Intel renewing its commitment to older product lines. The chip giant recently released the dual-core Pentium T3400, which Rain Recording used as the basis for its new Zephyr laptop.

"There's a slight sigh of relief to see that Intel is going to help me survive it too. The message is, 'It's okay to deploy mature technology.' Now I'm able to slice a significant amount off one of my mobile offerings," he said.

Like Paschick, Toste thinks Intel will be very conscious of what its channel partners need in the tough months ahead.

"They are cognizant of the channel's needs. To that point, in terms of their visibility to us, managing inventory, they're pretty good about it. The successful companies in these times are the ones that successfully manage the supply chain. It's no different for Intel," he said.

That said, the nature of Intel's business can make it extremely difficult to change course. Enormous amounts of money have already been spent developing designs and tooling factories to get Intel into the position it's in today. Now the company has "to try to get the market to catch up with their factory," Toste said.

Or as Peddie puts it: "The big ugly monster is the manufacturing fabs. It's a $2 billion to $5 billion commitment that's going to play out across five years. Once you pull the switch, there's no going back."

The thing is, economists are already talking about the recovery in terms of years. Otellini's rosy morning could still be a long, long way off. Could it be long enough for all of us to wake up that day looking at a very different version of "tick-tock" than the one we know now?