Analysis: Will ARM-Artisan merger be a defining moment?

The company is roughly one month away from finalizing its $1 billion sale to MPU vendor ARM Ltd. Thus far, all has been quiet as the regulatory process grinds on. The acquisition is on track to close as early as mid-December and isn't encountering any unforeseen roadblocks, financial analysts said Thursday (Nov. 18).

When the acquisition was announced in August, Wall Street and ARM investors applauded the technology play but voiced dissatisfaction with the price tag, which is roughly eight times Artisan's revenue.

Industry observers add that only time will tell whether or not the acquisition was worth the price.

They added that the acquisition's success will hinge on ARM's ability to leverage Artisan's presence in the commercial, off-the-shelf market. It will also depend on ARM's ability to sell more MPUs to foundry customers and library engagements to ARM's vertically-integrated customers.

At the same time, the companies need to convince the market that Artisan libraries will remain vendor neutral and that library quality won't be compromised by the acquisition.

The acquisition arguably combines ARM with the strongest standard cell libraries vendor — with links to the world's largest foundry — at a time when most companies are using foundries to produce silicon. Potentially, the merger could help the industry overcome major IC design roadblocks by combining ARM's systems business expertise with Artisan's understanding of the physics that could potentially slow silicon development.

"I think this acquisition will prove to be a defining moment that really helps establish IP as an industry," said Lucio Lanza, Artisan's chairman and principle of the venture capital firm LanzaTech ventures.

Artisan is no stranger to paradigm shifts in the IP business. Depending on who you talk to, Artisan either revolutionized or gelded the library business when it introduced its free licensing model for libraries along with Taiwan Semiconductor Manufacturing Co. Ltd. (TSMC) in the late 1990s.

Before Artisan's gambit, the library business was dominated by Aspec Technology Inc., which at its zenith received millions of dollars upfront from foundries and fabs to characterize their processes, and millions more from designers using the libraries for their next design.

Aspec offered general-purpose libraries that allowed customers to shop their designs to the lowest bidding foundry. Aspec's single library served several foundries, and its very broad rules base left performance and power on the table.

Artisan spotted this Achilles heel and countered with a "free libraries" royalty model with standard cell libraries optimized specifically for TSMC. In essence, the company offered libraries that were far better than competitive offerings, optimized to the world's largest foundry and came with no upfront costs. TSMC gave away Artisan libraries to customers and paid Artisan a royalty for every die built with the Artisan libraries. The strategy worked in part because designers who typically weren't TSMC customers would test-drive the libraries simply because they were free. The result was more TSMC engagements, which ultimately increased market share for Artisan and TSMC as many customers found big performance, accuracy and especially cost advantages.

Artisan's model spelled doom for Aspec and severely hobbled library startups Virtual Silicon and Duet Technologies Inc.

Those companies had to switch to the free libraries model and hope that royalties would eventually kick in. Virtual Silicon survived because it quickly secured a free libraries partnership with United Microelectronics Corp. (UMC) Duet, like Aspec, wasn't as lucky. The free library model also ended Avanti's foray into the libraries business, which it gained through the acquisition of Compass Design in 1997.

Meanwhile, Virage Logic has adapted to the free model change largely because it specialized in memory libraries, not the crowded standard cell business. Since then, Artisan and Virage have become more competitive by offering memory IP in addition to standard cell libraries. Virage is alos offering standard cells in addition to memory libraries.

Barry Hoberman, Virtual Silicon's president and CEO, said the company continues to offer free standard cell libraries to UMC, but said differentiation has proven to be the key selling point of today's libraries business.

Hoberman said general purpose libraries have become commodities, and foundries dislike paying royalties for commodities. Indeed, this is one of the reasons that Artisan, Virage and Virtual Silicon offer specialized libraries.

Virage offers low power and high performance libraries. Artisan offers similar specialized libraries, but it has expanded its focus beyond TSMC. This expansion, say industry observers, may have prompted TSMC to begin developing its own specialized libraries, thereby eliminating some royalty hassles.

Some observers speculate on whether it may also be the reason Artisan agreed to the $1 billion acquisition. "They've certainly set an attractive benchmark for valuations in this space," said Hoberman.

Jim Ensell, vice president of marketing at Virage Logic, agreed that the $1 billion price tag not only validates the physical libraries business but leaves Virage as the only large library vendor whose sole focus is on physical library development.

Brani Buric, Virage's senior director of product marketing, claims that gives Virage a distinct market advantage as new and ever more expensive processes come online. When that happens, foundries will want to work only with proven physical IP vendors.

"The cost of failure for the foundry and their customers is too high," said Burich.

Ensell said the merger's high price tag will put great pressure on the combined company to immediately show profitability "above what either would have done on their own."

ARM and Artisan executives said they could not discuss their strategy until the acquisition is completed.

However, ARM Chairman Sir Robin Saxby told EE Times that the acquisition stems from collaboration between the two companies on ARM's next-generation MPU core, code- named "Tiger."

While the ARM architecture has not always been the highest performance RISC architecture, Saxby said it has "always been the leader in the Mips-per-watt segment of RISC processors."

During the collaboration, it became evident that Artisan libraries would give ARM a better understanding of the intricacies of new silicon processes, Saxby said.

"As we get to deeper submicron technologies, all the physics gets a lot harder," said Saxby. "From a business point of view, now was the right time for this acquisition."

"Both companies offer IP that people really want, and the fundamental demand of both companies' business keeps renewing itself," added Artisan's Lanza, noting that these principles were overlooked by most IP companies.

The acquisition also brings with it a large market play. ARM has traditionally been strongest in vertically integrated design houses and semiconductor companies; Artisan's greatest appeal is with customera using a COT to foundry flow and has had close relationships with foundries.

"Looking to the future, the foundries play a greater role," said Saxby. "Artisan has historically had a close relationship to the foundries. As we hit the 90-nm node, the challenge for foundries and the semiconductor industry as a whole changes. If we can implement an architecture more efficiently with Artisan libraries and make it easier for partners to get to production and reduce the risk, we help everyone." Having ARM MPUs built with optimized Artisan libraries tailored to a foundry process essentially sweetens the proposition of using ARM cores over competitors' cores.

There is a growing list of competitors in the processor market, especially those in configurable processing like Tensillica and ARC. Saxby said ARM will keep Artisan's libraries open to all comers. "We have no intention of breaking Artisan's business model in the sense that they supply their libraries to our competitors," said Saxby. "This isn't about control, it is about enabling."

An example is ARM's AMBA on-chip bus, which is often used and targeted by competing core vendors. "The industry needs standards, connectivity and all the pieces of the [puzzle]," Saxby said.

Whether or not the design industry and foundries embrace the ARM-Artisan merger and establish it as the de facto standard standard library remains to be seen, observers said.

Virage's Ensell said the acquisition has "solidified" Virage's relationship with ARM competitors, MIPS, ARC and Tensillica. "They have approached us and want to work with us more energetically than they have in the past," said Ensell.

While business models have reshaped the industry, the best libraries nearly always wins, industry executives agree.

Saxby acknowledged that there's a lot of hard work ahead, adding that keeping Lanza and other Artisan executives engaged is critical to the success of the landmark merger.