Intel Eyeing Xilinx?

Christopher Danely, an analyst with J.P. Morgan Securities, recently speculated that Intel, which has about $13 billion in cash, will acquire Xilinx. A detailed analysis of his argument is available at EETimes. Hans Mosesmann of Raymond James Equity Research disagrees with this view, saying that Intel really wants to be relevant in wireless and smartphones, and that it has already tried its hand in programmable logic and sold its PLD business to Altera in 1994, so it is not likely to get into it again.

Meanwhile, China Mobile and China Unicom, two of the three major telecommunications companies in China, recently lowered their 2010 capital expenditure forecasts. This is bound to affect the programmable logic device (PLD) sector, which last year got a boost from the expansion of 3G mobile networks in China. Let’s take a closer look at the dominant players in this sector, Xilinx Inc. (NASDAQ:XLNX) and Altera Corp (NASDAQ:ALTR).

The build-out of 3G cellular network in China was a bright spot during the downturn. Analysts were expecting declines in 2010 as the majority of the build-outs are complete, but they were not expecting the declines to be so steep. China Unicom will cut its capex budget by 35% while China Mobile forecast that its capex would decline by 5%. Overall, Chinese carriers will reduce their expenditure by about 16% from an increase of 46% in 2009.

That is quite a blow to Altera and Xilinx, because the Chinese base station end-market accounts for about 8% of total Xilinx revenue and 12% of total Altera revenue. JPMorgan Chase maintained its Neutral ratings on both companies: “We believe Xilinx and Altera are at risk of an inventory build as their sales are roughly 10% above the previous peak while revenue at most of their customers should still be roughly 10% below the previous peak.”

Barclays Capital was more optimistic, noting that wired telecom business from Cisco, HuaWei, ZTE, and others remains strong. Barclays also expects capex for wireless systems in the United States to be up about 12% in 2010.

Prior to these announcements from the Chinese telcos, Altera had increased its first quarter revenue guidance to 7%-10% sequential growth. For the full year, Altera reported revenue of $1.2 billion, down 13%, and net income of $251.1 million, or $0.84 per share, versus net income of $359.7 million, or $1.18 per share, in 2008. Fourth quarter revenue was up 16% y-o-y and 27% q-o-q to $365 million, and net income improved to $103 million, or $0.34 per share. Earlier coverage is available here.

Gross margin improved to 68.4% from 67.3% in Q3. Altera ended 2009 with $1.5 billion in cash and short-term investments and declared a dividend of $0.05 per share. The stock is currently trading around $24 with market cap of about $7 billion after hitting a 52-week high of $25.75 on March 17.

Xilinx, the leader in the PLD market with annual revenue of $1.8 billion, reported third quarter revenue of $513.3 million, up 12% y-o-y and 24% q-o-q. Net income was $106.9 million or $0.38 per share versus $119.4 million or $0.44 per share last year and $64 million or $0.23 per share last quarter. Earlier coverage is available here.

Gross margin was 64.1% up from 61.9% last quarter. Cash and investments increased to $2 billion and the company ended the quarter with a net cash position of $1.3 billion. It announced a quarterly cash dividend of $0.16 per share.

For the fourth quarter, Xilinx expects revenue to be up 3% sequentially to down 1% sequentially. Gross margin is expected to be approximately 64% to 65%. The stock is trading around $26 with market cap of about $7 billion after hitting a 52-week high of $26.89 on March 5.

My perspective is that programmability in chip design has become a key factor that is driving a very large portion of the semiconductor business at the moment. The pricing pressure on electronics of all kinds, especially consumer, computers and communication gadgets, is extremely high, making the pressure on the chip ecosystem intense.

Chips, therefore, have to be designed for less money but with more functionality, which makes PLDs a better strategy than ASICs and ASSPs, especially for mid-volume products. Consequently, it would be reasonable to say that all but the highest volume chips are going to move to some form of a programmable solution just to keep up with the business model changes.

In 1994, when Intel sold its PLD business to Altera, this was not the case. ASICs and ASSPs grew aggressively, fueling an electronics renaissance for a decade. But times have changed. Most important, business models have changed to the extent that the semiconductor business is fast becoming a non-profit industry. The pricing and margin pressure on electronics will continue over the next decade as emerging markets ramp up.

Against this backdrop, PLDs look to be a relatively viable business model, and I would not be surprised to see Intel eyeing Xilinx.