Inside Intel’s numbers: What the capex surge means

by James Montgomery, news editor

January 17, 2010 – Historically countercyclical-investing Intel is up to its old tricks, pledging a massive record spending spree in 2011. We break down all the numbers and what they mean, from exec statements and Q&A comments to what key industry analysts think.

Key data points from Intel’s 4Q10 and full-year 2010:

  • 4Q10: Revenues $11.5B, up 3% Q/Q. Gross margins 67.5%, up 1.5 points Q/Q. Net income $3.4B (up 15% Q/Q), EPS $0.59 up 13% Q/Q. Revenue, net income (and operating income), and EPS were all records.
  • By business, data center group was up 15%; other businesses were flat (PC/client, other architecture, Atom and chipsets). MPU ASIs were up slightly sequentially GMs and R&D+MG&A spending ($3.4B) were both higher than Intel’s expectations.
  • 2010: Revenues $43.6B, up 24% from 2009. GMs 66%, up 10 points. Net income $11.7B, up a whopping 167% Y/Y, and EPS $2.05 up 166%. Those include fines from the EC ($1.45B) and settlement with AMD ($1.25B) and related tax impacts.
  • PC client sales were up 21%, data center 35%, other architecture 27%, Atom and chipset 8% (half of Atom design wins were in China).
  • Full year capex was $5.2B, in line with expectations. Thanks to equipment reuse from 45nm manufacturing, Intel is offsetting almost half of the required capital investment at 32nm, according to CFO Stacy Smith.

Key figures for Intel’s 1Q11 and 2011 projections:

  • 1Q11: Sales flat at $11.5B (±$400M). GMs down to 63%-65%.
  • FY11: Capex $8.7B-$9.3B, up around 73% from 2010. Total spending (R&D plus MG&A) $13.9B, up about 8%. GMs 65% with "a few percentage points" wiggle room.

Comments from the Q&A:

Revenue expectations, emerging markets: Paul Otellini expressed "confidence that we can grow our revenues by approximately 20% in 2011," without taking into account the Infineon wireless or McAfee businesses. Stacy Smith quickly chimed in that 10% revenue growth vs. 2010 "is the expectation that we’re setting." That assumes, he clarified, a "low to mid-teens" growth in PC units. Emerging markets are where the growth is coming from, 2:1 vs. "mature" markets (read: PC). And in China, PCs cost roughly seven weeks of income, making it affordable to more than a billion people. Another big growth area: servers to help build out the cloud computing infrastructure.

22nm technology yields are being ramped, design of the first MPU is completed and working devices are created. Production wafers will be ramped in 2H11, initially for mainstream microprocessors, but then "rapidly" moving to "the non-PC part of the product line," said Otellini.

A projected low- to mid-teens growth in 2011 unit growth, but 10% sales growth, implies ASPs will decline, pointed out one caller. Smith attributed this to expectations of growth in consumer and emerging markets, weighing the mix to come down, a trend that he’s "betting on for the next several years." That unit growth also implies "a pretty robust corporate refresh cycle continuing in 2011," Otellini said.

What’s driving $9B capex: That $9B in capex, for >70% Y/Y growth — that’s true to Intel’s countercyclical investment history, as industry watchers expect overall 2011 to be flat or slightly up this year. Most of that $9B capex will go to equipment (vs. fab depreciation), but "you’ll see a little more factory spending than what you’ve seen in the past," Smith noted. (Investors subsequently boosted key equipment supplier stocks, including AMAT, NVLS, KLAC, and LRCX.) Depreciation will rise from $2.4B in 2010 to $5B in 2011. PCs and servers are driving that $9B in capex. Things like smartphones and tablets are small-die, but it’s also strategically important to get those products into 22nm helps bring down costs and improve performance. Smith: "Our investment profile increased a lot in [2010]. It’s increasing more in [2011]. We’ll make the decisions for ’12 when we get there."

Intel also is shifting to a four high-volume factory model vs. three due to unit growth over the past couple of years, and the company’s aggressive transition to leading-edge process technology. "We do that, because we get paid for that process technology leadership. We get paid in terms of having lower costs," Smith said, pointing to the transition of graphics processors, as well as the aforementioned savings in 32nm equipment. ("We are interested in seeing how Intel’s long-term gross margins are affected by this move to bring graphics functionality up the N process node," wrote Craig Berger of FBR Research in a research note.)

Analysts’ take: Intel’s capex splurge

Intel’s capex vs. revenues will rise to 19% in 2011, vs. 12% in 2010 and around 13% the past three years, reflecting less reuse at 22nm than 32nm and anticipation of more contributions in noncore markets by 2012, notes Credit Suisse’s Satya Kumar.

Doing some extrapolation and math, Kumar sheds an especially positive light on Intel’s lithography spend — the company previously said it would increase its litho spending to 45% of total capex at 22nm (vs. 35% at 32nm). Assuming a 65% WFE/capex ratio out of the "lion’s share" the company said would encompass that $9B 2010 figure (and 80% of it on 22nm, vs. 20% in 2010), Kumar arrives at a $2.7B litho capex spend, more than twice the $1.25B in 2010. Big winner here: ASML, which gained share from Nikon at this key account for 22nm (65% share, double its 32nm share). He translates that to $1.3B sales for ASML from Intel, up from $400M in 2010, and a possible $500M revenue upside.

Intel’s assumption of 10% growth in 2011 without much market-driver oomph from "emerging" tablet and smartphone markets, if extended to 2012 as well, leads Kumar to believe more greenfield fabs will be needed by foundries, and soon — given nearly two-year schedules to ramp a greenfield facility from inception to chip output — and increased complexity at 22nm from design to mask levels to yield ramps. GlobalFoundries has barely started staffing its new fab in NY and doesn’t expect pilot production until 2012, and "meaningful volumes" not until 1H13, he notes. That longer time-to-output-capacity for greenfield fabs "could suppress additional
productive capacities coming online in 2011 and 2012," he notes.

A rising capex tide won’t lift every boat. Besides ASML, Kumar sees good news for KLAC, NVLS, CYMI, and AMAT — but not LRCX, which has less exposure at INTC. VSEA should also benefit, though he thinks Intel’s use of implant will be lower at 22nm than at 32nm.


Easily post a comment below using your Linkedin, Twitter, Google or Facebook account. Comments won't automatically be posted to your social media accounts unless you select to share.