Spansion/Intel/STM: Phase Change Phantoms

Observers of the three-way fight between Intel (INTC), Spansion (SPSN) and the DRAM industry have got to admire the move made Thursday by the Intel/STMicro (STM) team, aka Numonyx.

In announcing "the most significant non-volatile memory advancement in 40 years," Ed Doller, soon-to-be CTO of Numonyx, seemed to be signaling that the third place team is not just coming back but about to take over.

The phase change chip is "sampling" as a drop-in replacement for NOR, with the PCM technology itself having "the bit-alterability of DRAM, the non-volatility of flash, the fast reads of NOR and the fast writes of NAND." Last March, Doller suggested mass production in 2H, 2007, and was quoted as having said "Phase change technology gets us pretty close to nirvana."

But one has to be skeptical of a memory paradise so sunny that Intel is aching to leave it. By spinning Numonyx off and taking cash out, Intel is leaving paradise, and leaving it under-funded. Why not just buy out STM's NOR business, cut overhead and keep paradise in-house?

And, if PCM is truly a game changer, why launch it in the backwater puddles of the flash market? What is Intel's rational for passing on DRAM? And on NAND - a market Intel likes, which is 3x bigger than NOR and growing much faster?

With a 20% position in SPSN, I admit Thursday's release, like Intel's NAND "break-through" a few days ago, was initially disconcerting. But the fact is, everyone is working on new flash technologies. Both Samsung and Spansion are involved in the new Onfi standrads behind Intel's NAND play, and everyone is saying they've got the latest and greatest.

But, behind the talk, there is one company with a cutting-edge NOR model in handsets today, and there is another with a Hail Mary chip that might, by its own admission, be cost-competitive when the industry gets to the 22nm process node. Which would you rather own?

Disclosure: Author has a long position in SPSN

Comments

[David Lentz]

Perhaps there are other technologies in Intel's labs -- beyond flash, DRAM and PCM -- such that they can see that a large investment in PCM would be wasted effort, as it is likely to be overturned itself at about the time that the investment would begin to pay off. Something like memory based on quantum-dot technology (or pick the "out-there" buzzword of your own choosing).

At a time when technologies are changing rapidly, and the investments in time, effort, and capital needed to bring them to market grow ever larger, a number of opportunities are perhaps better exploited by putting the minimum necessary into a venture to give it a shot at survival, and capitalize it via an IPO.

None of this means that PCM is not a technology worth investing in, or that it will not push flash memory aside, but only that in Intel's opinion (and presumably they are better informed that we are on such matters), the best use of their time and capital is in other areas, and that the best way for them to benefit fom the PCM business is to spin it off and allow it to grow using the capital resources of the marketplace, and their own time and effort.

The goal here is the most prudent management of the shareholder's money, not the domination of any particular segment of the semiconductor marketplace.

[nvm expert]

PCM Overturned? Hmmm... lets see... DRAM, EPROM --> Flash, and SRAM have been around for at least 30yrs. The fact is the ability to scale these technologies is causing the lage investment in the industry. Intel and hence Numonyx (Intel / ST spinout) will be in a great position should they be able to drive the technology into the industry.

The memory market is >$30B. Having a substantial lead in a technology that has the potential to displace the current technologies seems like a prudent investment to me.