450mm plus EUV: has Moore’s Law met its Waterloo?

Intel cemented the future of 450-millimeter wafer technology with a recent $4.1 billion investment in ASML, a leading provider of lithography machines for the semiconductor industry. The move helps share the cost of developing bigger silicon wafers that yield more chips. ASML was seeking equity partners to not only migrate from the current 300mm wafer standard, but also finance research for Extreme Ultraviolet technology, or EUV lithography. Intel aims to secure a 450mm prototype by 2015 and push next generation silicon hard enough to shake TSMC and GlobalFoundaries, its remaining rivals in the ever-expensive game of chip manufacturing.

Both Intel and ASML argue the bigger wafer deployed with EUV technology is the industry’s best chance of keeping pace with Moore’s Law. Deeply rooted in Intel’s culture, Moore’s Law means doubling the number of transistors on a silicon chip every two years. It’s also the law of the semiconductor jungle, a market expectation, and bastardized as an opportune marketing tool for loosely related industries. Yet people forget Gordon Moore was actually the co-founder of Intel. And his transistor prophecy penned in 1965 still drives semiconductor innovation whether the competition likes it or not. And almost 50 years later, Intel isn’t backing off the gas.

"This funding and participation by a leading semiconductor manufacturer is an acknowledgement of the essential contribution of lithography technology in ensuring the continuation of Moore's Law,” Eric Meurice, chief executive officer of ASML.

According to ASML, keeping up requires making lithography machines that allow chipmakers like Intel to pack transistors tighter with each generation of manufacturing. Print finer features. Shrink the size of transistors. This tricky task boosts processing speed, lowers power consumption, and increases efficiencies throughout our world of consumer and enterprise computing devices. There’s more to Moore’s Law, but Intel is concerned about squeezing the life out of 300mm wafers and ASML may hold the keys to the 450mm kingdom. At least that’s how Intel sees it.

“By moving from today's standard 300mm wafers to new larger 450mm wafers, the industry can effectively double the capacity of its factories for only a fraction of the cost,” said Brian Krzanich, Intel senior vice president and chief operating officer.

Intel is the big, hungry lone wolf of the semiconductor biz. It’s the only truly integrated manufacturer left, designing and producing chips for an owned and branded portfolio. Even IBM and Samsung have joined forces with GlobalFoundaries to reduce the risks and costs of R&D.

Intel also hunts in many markets. And as a byproduct of their integrator status, have competition at multiple levels of the semiconductor food chain. Beyond PCs and storage, Intel now wants to eat in the mobile space and is thus tangling with a different breed of cat. So we have Intel’s classic competition in the foundries and AMD, and increasing pressure from Qualcomm, Samsung and Texas Instruments, all growing based on 2011 numbers. Throw in gaming specialist NVIDIA and Intel could be vulnerable protecting so much territory.

But it’s not going down like that. Intel is still the alpha chip.

According to IHS, Intel has more market share and total revenue. Revenues for 2011 were up 20.6%. And after the ASML deal was announced, analysts placed a strong buy rating on Intel’s stock. Intel has long considered manufacturing and production a differentiator and sustainable advantage, especially in a process-intensive business requiring billions to outfit a factory with leading-edge equipment. In this case, retooling to build a 50-percent wider wafer etched by EUV technology. It’s the kind of power play Intel can afford, and all in the name of Moore’s Law.

In today’s broader tech world Intel doesn’t grab the same headlines saved for Apple and Google. And we take faster, meaner processors for granted, leaving Intel with a grandfatherly image and a commoditized reputation. But closing these now assumed, next-gen gaps in production is fine science and commercializing EUV technology is no exception. It takes both brains and brawn. Something Intel has - and ASML needs more of - to get to the right side of the semiconductor roadmap.

So all bets of keeping pace with Moore’s Law seem to ride on EUV. Yet EUV has proven a difficult proposition. Lithography is an optical technology, and ASML is busy trying to shorten the wavelength used in the process from the current 193-nanometer ultraviolet light to 13.5 nanometers of EUV light. Imagine something between X-ray and visible light on the spectrum. ASML maintains this shorter, powerful, and more optimum wavelength creates a higher resolution and smaller features.

All forms of matter including air molecules absorb light at these diminutive EUV wavelengths, and so exposure must occur in a vacuum chamber. A bright enough light needs to reach each targeted wafer for the process to work. Losing light to air molecules - or anything else - is ASML’s challenge. And although current ASML prototypes are billed as EUV technology, they have throughput problems. The prototype beams aren’t powerful enough to reach desired economies of scale.

Intel helped form the EUV consortium back in 1997 and wanted its current line of processors, called Ivy Bridge, to feature EUV technology. So EUV is not new, but it is troublesome. ASML wants to throw more R&D money at it and keep cooking. They are seeking shared investments beyond Intel to keep pushing forward.

The competition is not sitting idly by waiting for ASML to bust through with the 450mm and EUV combo. TSMC announced in June an $8-10 billion investment in similar technology. And like ASML, TSMC doesn’t sweat the 450mm barrier as much as developing the technology to reach production levels and deliver a worthwhile payback. At this point, ASML seems to have the best plan and the industry is leveraging more ways to share costs. In fact, both TSMC and Samsung are in negotiations with ASML to take part in the same equity and research program Intel has jumped on.

So is the semiconductor business simply at a technological crossroads, or is 450mm plus EUV the Moore’s Law Waterloo?

Not everyone in the semiconductor business is crazy about this revolutionary chip concept. Or sold on, based on today’s market dynamics and manufacturing costs, the importance of doubling the number of transistors on a silicon chip every two years. But respecting Intel’s history and bankroll, it’s hard to bet against the Moore’s Law rallying cry. And the ASML co-investment, coupled with an aggressive move toward 450mm adoption is probably not rooted in the best interest of the industry. More like Intel flashing their teeth and attacking the future, the alpha chip strategy for separating farther from the pack and leaving weaker players behind.