Atmel designs, develops, manufactures, and sells a wide range of semiconductor IC products, including microcontrollers, advanced logic, mixed-signal, and nonvolatile memory and radio frequency components, with an emphasis on complete system solutions incorporating microcontrollers. And 2007 was a very good year.
After placing 99th on our 2006 list, Atmel charged up the charts to take the 56th slot last year despite a slight decline in sales. This surge came from improvements in margin and in operating profit after two years of losses. The company’s troubles began a while ago, with too much of a buildup in fabrication capacity and a huge investment in its loss-ridden NOR flash business, leading to a period of restructuring and turbulence.
Now, Atmel is focusing on enhancing its mix of microcontroller products, as that’s where the high margins are. It also is continuing to reduce its exposure to expensive fab sites and rationalizing product development toward its best opportunities. It saw some changes at the top as well, as Atmel fired president and CEO George Perlegos and three other officers over alleged misuse of corporate travel funds in 2006. Stephen Laub, a director, replaced Perlegos, who failed in a 2007 attempt to regain control of the company through a long and bitter proxy battle.
Atmel’s move toward a “fab lite” strategy started back in 2005, when it sold its fab in Nantes, France, to XbyBus, a French-based components supplier. In July 2006, it sold its subsidiary in Grenoble, France, to e2v Technologies plc. That deal also included a wafer fab. In May 2007, Atmel sold a wafer fab in Irving, Texas, to Maxim Integrated Products. And in July 2007, it sold its network storage unit to MoSys Inc.
In October 2007, Atmel agreed to sell its 8-in. wafer fab equipment from its fab in North Tyneside, England, to Taiwan Semiconductor with the land and buildings sold to Highbridge Business Park Ltd. This move should reduce Atmel’s debt by $35 million. It also should reduce future capital expenditures and improve capacity utilization in other facilities. Atmel is still looking for a buyer for its fab in Germany.
Why are companies like Atmel and Texas Instruments going “fab light”? Besides lowering capital expenditure requirements, a company that contracts out a portion of its manufacturing may be able to bring that piece back in-house in the event of an economic slowdown. This helps maintain margins, which is a big advantage in a cyclical, capital-intensive industry. Companies like Taiwan Semi and United Micro also benefit as they exist for the purpose of third-party outsourcing. A modern chip fab costs several billion dollars, so it’s more effective to use third parties.
In August 2007, Atmel announced plans for a $250 million share buyback, as much as 10% of outstanding shares. It certainly had enough cash on the balance sheet to use for the purchase. Atmel’s thinking here may be that management is for seeing a turnaround in its business for 2008 and beyond due to its restructuring and current focus and wants to buy shares before the rest of the market wakes up to Atmel’s potential.
In focusing on high-margin microcontrollers (about 30% of its sales), Atmel has sold its WiMAX development group in Germany and may sell other product groups with weaker margins or slower growth. Atmel also has signed an agreement to acquire Quantum Research Group, a developer of capacity-sensing intellectual property and solutions for user interfaces. Quantum’s sensors usually are sold with a microcontroller, so Atmel can expand its microcontroller activities further into the capacitive sensor market.
Laub continues to restructure and reduce headcount strategically, as there is a real opportunity to increase margins by reducing the company’s European manufacturing presence. This will most likely be a two- to three-year process, as products moved to different facilities need to be requalified. The aforementioned German fab in Heilbronn and a fab in Rousset, France, both have high costs and pension expenses compared to Asian fabs.
Atmel could attempt to be come a pure microcontroller player over the long term. About 30% of Atmel’s sales are in this high-margin segment. These chips have long product cycles. They can provide a revenue “annuity” model, so to speak. And, they’re very broad-based products that can be sold to thousands of customers. Focusing on non-commodity, high-margin products makes a lot of sense.
The huge problem with this approach may be that the other 70% of Atmel’s sales are in commodity or other low-value parts. Shifting away strategically from such a large percentage of one’s sales base can become a very long and painful process.