by James Montgomery, News Editor, Solid State Technology
Feb. 12, 2008 – Industry watchers tell WaferNEWS what likely spurred the unsolicited proposal from Sumitomo to acquire JV partner Axcelis; what’s good and bad about such a combination; and how might a changed implant sector landscape be positioned for the next key period: 22nm and vertical transistors.
After what it claims was nearly two years of rebuffed approaches, Sumitomo this week launched an unusually public proposal to acquire Axcelis Technologies, its 50/50 JV partner in Japan (Sumitomo Eaton Nova, or SEN). For now, Sumitomo is waiting for an answer, while Axcelis execs say they’re reviewing it (but no timeline provided).
One of the first questions to ask is, how did this happen? Public discussions by Axcelis execs (notably quarterly conference calls) have been peppered with hints of discord for roughly two years, and even participating analysts have continuously asked about how the relationship stands.
Industry sources contacted by WaferNEWS — Dean Freeman, research VP at Gartner, and John. O. Borland, SST editorial board advisor — seemed to agree that SEN actually was faster to recognize the need for 300mm tools, and Borland pointed to SEN’s endstation (specifically requested by Taiwanese customers) with a gyroscopic design that Axcelis eventually adopted.
But perhaps the final straw concerned Axcelis’ refusal to license its Optima Imax high dose, low energy implant tool to SEN in Japan — and in the company’s latest quarterly call, execs indicated the company plans to sell the tools direct in Japan, recording sales in 2008.
“Until around 2005, Varian [Semi. Equip. Assoc.] had very little traction in Japan,” Freeman noted, but single-wafer tool stumbles by Applied and lateness by Sumitomo/Axcelis allowed Varian to “walk in and pick up Toshiba and Elpida[‘s]” high-current business, and now can use that as a bridge to sell their medium-current offerings. Axcelis is just now “beginning to make enough inroads to pick up […] people late to 65nm and below,” he said.
Meanwhile, market dynamics also are putting pressure on a tighter relationship. The industry can’t support 3-5 suppliers in each sector anymore, Freeman noted, and now it’s only two or maybe three: “the ‘process of record’ and a competitor to keep prices down,” he said. And right now, “Axcelis is the second vendor [customers use] to leverage Varian over the next few years.”
Speculating if the deal is approved, both Freeman and Borland acknowledged several questions remain, including the proposed management structure — to what extent Sumitomo would take a dominant role, how they would (re)structure their Japan JV operation, how Japan’s Sumitomo would manage US-based Axcelis (e.g. with domestic leadership or sending Japan execs overseas), etc.
But the main concern facing Axcelis and SEN/Sumitomo would be whether, as a combined and streamlined entity, they can push forward with best-of-breed technology to regain share against market leader Varian. Axcelis/SEN “had an extremely loyal customer base,” up until those customers needed the single-wafer high-current implant technology that the two suppliers couldn’t yet provide, Freeman noted. “Can they win that customer base back? Or because of their failure to execute, have they lost it forever – or until Varian trips?”
The answer may come at the next inflection point for this sector, around the 32nm-22nm node (logic), and the introduction of vertical transistors, where “implant is very critical,” Freeman noted. Chipmakers will have to choose between plasma or beamline implant, and the game will be who has the better tool platform for it.
In terms of beamline, Borland thinks that, right now, Axcelis may have a leg up with its spot-beam technology vs. Varian’s ribbon broadbeam technology, with nonuniformity the key differentiator. In the next three years or so (~22nm), he told WaferNEWS, chipmakers will have a choice: “either switch vendors or switch dopant species” (currently phosphide, which offers better activation than arsenic, and maybe later antimony). Similar to their choice to move from batch to single-wafer platforms, it’s going to be a decision based on device yields, and that’s the only way he sees a platform switch taking place.
Beamline aside, Axcelis/SEN may have an advantage in plasma as well. Borland pointed out that SEN sells a plasma implant tool in Japan’s TFT market, so it could apply that knowledge to similar technology for semiconductors. “It requires much more control, and means adding hardware and software, but the basic concept is the same,” he sad.
Freeman added that a formal Axcelis/SEN combination may not be the end of this story, noting that economies-of-scale in this industry now means companies really need ~$200M in annual sales to even play in niche markets/regions, and more like $500M-$1B to be a significant global player. The question is, would ACLS/SEN be satisfied as a niche player, or would it want to grow bigger — and do it organically or inorganically? The latter opens up intriguing growth possibilities, he noted, speculating future attractive combinations with oft-rumored industry consolidators like TEL, NVLS, or LAM — or even ASMI, whose shareholders continue to push for a split of its frontend and backend businesses, and which lacks a presence in doping, Freeman pointed out.
So it seems there are many reasons for Axcelis and Sumitomo to formally join forces in the implant sector. But perhaps the biggest roadblock is as yet unknown. A third industry source, not tied to either company, pointed out that Sumitomo’s actions are very atypical of Japanese companies, who traditionally avoid not only international M&A but also such public controversy (though also acknowledged the changing business environment where shareholder value is paramount, and foreign capital is increasingly making market inroads). Still, such a public hostile act against one’s 20+year JV partner “must mean things are really bad,” the source suggested, “and it’s hard to see how this is going to make them better.” J.M.