Portomaso, Malta — Are the wheels coming off the electronics industry's disaggregation train?

Roughly 20 years into the shift from vertical to horizontal integration, the mere question may seem heresy. But some executives who gathered here last week for the International Electronics Forum suggested that, as Synopsys Inc. chairman and CEO Aart de Geus put it, "we are at one of those switching moments again."

Few would deny the importance of maintaining some level of control over the system-on-chip design process. The question is whether sufficient control is possible under the disaggregation model.

De Geus, for one, argued that SoC complexities and the pressure of "time to results" in the now-dominant consumer market demand a more "systemic approach" to design.

And at least two Japanese consumer giants are nudging the pendulum back to some level of vertical control. Sony Corp. and Matsushita Electric Industrial Co. both used the forum to unveil plans to strengthen their internal semiconductor businesses.

Tsugio Makimoto, a Sony corporate adviser, said the company, building on the bulwark of the ambitious Cell processor it devised for the third-generation Playstation, plans to increase the proportion of internally developed chips in its systems from 20 percent of the total chip content in value terms to 40 percent over the next three to four years. Sony annually consumes $8 billion worth of chips, or 4 percent of the worldwide semiconductor market.

Osamu Nishijima, executive officer and vice president responsible for Matsushita's semiconductor business, revealed his company's plans to roll an internally developed scalable silicon architecture, called the Integrated Platform, for use in Panasonic-brand products ranging from digital televisions and DVD recorders to digital cameras, MP3 players and cell phones.

Defying skeptics within Matsushita's own system product divisions, Nishijima said the architecture, which was five years in the making, has proved itself capable of being "optimized, with no compromise" for a diverse range of end products. System designers, he said, retain control over such decisions as whether to add another hardware engine to a given product or to exclude a parallel data processor from the design. The Integrated Platform, he said, ensures that the "software deployed in each product looks exactly the same to the system, regardless of choices made by system designers."

Neither Sony nor Matsushita has designs on an increased role in the merchant semiconductor market. Rather, the goal is to protect their market positions in consumer gear by tapping internal resources to differentiate their products from those being turned out in increasing numbers by Asia's ODMs.

Makimoto called it a "back to customization" movement. When the PC was the dominant market driver, he said, standardization around the Wintel platform and the rise of the fabless chip company opened the door to the horizontal business model, with design divvied up among a host of narrowly focused developers.

But PC revenue on the Japanese market, Makimoto said, now trails the sales number for the "big three" consumer platforms — TVs, DVDs and digital still cameras. In the relatively lawless world of consumer electronics, where no one platform dominates, some degree of reaggregation may make sense for established manufacturers with sufficiently deep pockets.

"I think the trend for disaggregation of the chain — breaking the supply link or the acceleration to go fabless — was driven by wrong reasons," said Malcolm Penn, chairman and CEO of Future Horizons, which hosted the forum. "If you are big enough, under a vertical-integration model, you are better-positioned for peaks and troughs because all of the money stays inside the company."

Moreover, the argument that an internal semiconductor business can't compete with merchant chip vendors is obsolete, Penn said. Digital consumer products are highly hit or miss, so an internal chip division may have an edge over merchant chip companies. "At least, they are closer to the customer," Penn said. Again, the issue is one of control over the process.

ARM Holdings plc chairman Robin Saxby agreed. "I just came back from Japan three weeks ago," Saxby said. "I can tell you that Japanese companies are getting much more focused, efficient and aggressive — and realistic."

But integration will never be as vertical as it once was, Sony's Makimoto cautioned. "Maybe the combination of horizontal and vertical models is the way to go."

Or maybe the pendulum needn't swing back at all, some sources said.

Walden C. Rhines, chairman and CEO of Mentor Graphics Corp., took issue with the view of his counterpart at Synopsys. "The industry shifted from a vertically integrated model to horizontal because of its needs for specialization and reduction of cost. Once that happens, an industry rarely reverses itself," Rhines said.

Indeed, he said, increasing numbers of system companies are shifting so far in the other direction that they're adopting the Dell model, in effect becoming system integrators.

Clayton Christensen, Harvard professor and author of the books The Innovator's Dilemma and The Innovator's Solution, sees "disruptions," rather than reaggregation, causing shifts in the design chain.

For example, the ability to design a microprocessor today is limited to highly skilled engineers. But as technology evolves, tools are emerging that allow software engineers, for example, to have a hand in processor design, vastly expanding that universe, he said.

"That's disruption. Whoever brings that ability, there's a big growth opportunity there, even though commoditization threatens growth on the other side," Christensen said. And those disruptive models are likely to bubble up in disaggregated form.

Meanwhile, the strains that pulled at vertically integrated companies in the past haven't gone away. Semiconductor units may be underfunded and, because they serve the parents' product groups, their cycles must satisfy the requirements of those divisions. Thus, they may face problems forming and executing long-range plans.

The latter point has hit home for NEC Electronics Corp., which has a high-profile presence in the Japanese mobile-phone market. "Our system LSI business needs to find a way to catch up with a six-month model-change cycle of cell phones," said Hiro Hashimoto, executive vice president at NEC Electronics.

Andrea Cuomo, STMicroelectronics' executive vice president and chief strategy and technical officer, noted that in a slow market, chip companies need to offer complete kits and solutions. Second-tier customers, for example, "no longer can afford their own software development," so the task falls to ST. The challenge, he said, is to manage the various "business models in parallel [and] to make money on every business model."

Sumita Sadana, senior vice president for strategy and business development at Freescale Semiconductor Inc., echoed the sentiment that "slowing growth, higher costs and [higher] complexity" in the industry at large are expanding chip vendors' focus from "silicon needs of customers" to the "complete customer value chain." The investment is no longer limited to silicon but includes tools, systems and software, he said. Thus, partnerships are the norm, and more aggressive mergers and acquisitions are likely on the horizon.

ARM's Saxby noted that even intellectual-property companies need to do "controlled business experiments" to find products and markets that are a good fit. ARM's partnership with Apple on the development of the Newton PDA, for example, greased the wheels of the ARM architecture's profitable pairing with iPod.

What does all of this mean for ordinary chip companies with no special partnerships or ties to big, powerful system vendors? "They are condemned to a commodity market," predicted Penn.

— Additional reporting by Ron Wilson and Brian Fuller