Profitless Prosperity Haunts IC Industry

Profitless Prosperity Haunts IC Industry

Alexander E. Braun, Senior Editor -- Semiconductor International, 1/17/2008 http://www.semiconductor.net/index.asp?layout=articlePrint&articleID=CA6523410

Profit levels in the semiconductor industry have dropped to alarmingly low levels, said Steve Newberry, CEO at Lam Research Corp. (Fremont, Calif.). For more than half of the top 40 semiconductor companies, current levels of profitability are “minimally sufficient” to fund new investments, he said in a presentation at the SEMI Industry Strategy Symposium (ISS 2008), being held this week at Half Moon Bay, Calif.


Steve Newberry, CEO, Lam Research Corp.
“Although we’ve experienced unprecedented year-over-year IC unit growth, we’re trending toward an era of profitless prosperity,” he said, defined as an industry running flat out without generating sufficient return on investment (ROI) to build new facilities necessary to meet growing customer demand.

“We better figure out how to achieve pricing that enables sufficient profit, because there won’t be enough banks or fools that will continue loaning money to companies that cannot generate what they need to fund the business,” he said, adding that consolidation is necessary to improve the situation.

In 2004, 38 out of the 40 largest companies had positive operating profits; the industry average was 23%. Even then, Newberry said, two companies had negative operating profits, and 15 were between zero and 10, which is insufficient to sustain growth. For 2007, the average operating profit is 12%.

“We’re in a strong demand environment, and 15 of 40 companies have negative operating profits,” he said. “If you add the eight that are from 10 to zero, that’s 23; over half the companies are under the profit level needed to sustain business. If you exclude known profit leaders Intel Corp., Texas Instruments Inc., TSMC, the fabless companies and analog, there’s a -1% operating profit. Only nine companies have positive operating profits.”

Since January 2003, the industry’s IC unit yearly demand has grown faster than during the previous 12 years. “IC unit demand is growing, but revenue growth is lagging,” Newberry said. Since 1995, average selling prices (ASPs) have dropped too fast for the industry’s own good. The price-per-bit, per function, has declined faster than the cost-per-bit and function, he said.

From 1992 to 1997 demand rose sharply and ASPs grew significantly, then plunged. “Even so, the industry’s operating profit was strong enough at 20% to generate sufficient cash to self-fund business growth,” he said.

Over the past five years unit growth has been 14%, almost double the average IC unit growth rate from 1998 to 2002. Profitability shot up in 2004, and continued to do well in 2005 and 2006 in the face of declining ASPs caused by shortages in some rapidly growing segments.

However, for 2007, the industry’s average profitability is forecast to be ~12%. “This is minimally sufficient, overall, for the industry to self-fund,” he said.

Newberry said that one cannot be prosperous if the price-per-bit declines faster than the cost-per-bit, which historically has averaged ~30%. “Today, we have an even greater accelerated price-per-bit decline, and have a huge gap between price and cost. Memory manufacturers are in a death race to the bottom.”

Newberry indicated that over the past few years profits for advanced logic IDMs, DRAM and NAND have fallen from ~20% to ~4%; the NOR sector has been below positive operating profits for five years.

In memory, oversupply causes ASPs to degrade faster than costs drop. Too many players have easy access to capital and are able to quickly ramp production. As a result, supply increases faster than demand, even with 50% annual unit growth. Memory companies seek the perceived benefits of economies of scale and increased revenue from market-share gains, Newberry said.


Outside of a few healthy companies, many chip companies have insufficient profits.

For logic IDMs, Newberry said only one logic IDM has an operating profit greater than 10% — Texas Instruments. “Even so, TI, with an operating profit in the 20s, has again changed its business model. They went fab-lite and are now going fab-lighter. They increasingly resemble a fabless company.”

The foundry and fabless segments are over 20%: one is capital-intensive, the other R&D-intensive. The foundry segment seems healthy, but over the past five years, Taiwan Semiconductor Manufacturing Co. Ltd. (TSMC, Hsinchu, Taiwan) has profited. UMC (Hsinchu, Taiwan), Semiconductor Manufacturing International Corp. (SMIC, Shanghai) and Chartered Semiconductor Manufacturing Ltd. (Singapore) have not.

“Cost isn’t the significant factor,” Newberry said. “TSMC gets a value leader price premium because of its library, product breadth and capability to deliver.” The microprocessor market looks good, with only two players. “Intel is eight times AMD’s size. So AMD’s strategy was to come up with a better processor,” Newberry said. “It worked until Intel price-nuked them. Now, AMD’s 2007 profitability is in the red.”

Newberry believes the industry must do more than focus on cost reduction. “We must promote market efficiency, global industry exchanges, extensive supply-and-demand inventory tracking, and practice financial discipline. We must develop better models to produce demand vs. supply forecasts. There’s too much capability chasing too little demand.”





^O^