Intel is up against it with its shares falling by a third in the last three months, with the enterprise computing market expected to soften, and with NAND flash prices falling like a stone just at the moment when it is in the most expensive phase in any chip sector i.e. when it is ramping up fabs to get into a new market.

Recently Goldman Sachs took Intel off its 'conviction buy' list for shares because it believes companies will buy less PCs. Then, a Goldman Sachs analyst suggested that: 'Intel will need to better articulate the reason behind its position in NAND before investors feel comfortable buying the stock'.

The cost of bringing up the NAND flash fabs in its joint venture with Micron called IM Flash, is so high that Intel blamed its NAND business for knocking a couple of percentage points off its gross margin forecast, even though the NAND business accounts for only 3 per cent of Intel's revenues.

While analysts are predicting a 55 per cent drop in NAND flash prices this year, Intel is saying the prices dropped 53 per cent between Q4 and Q1 alone.

Meanwhile, Toshiba is betting the entire company on building NAND flash fabs with the aim of becoming the number one player. At the moment Samsung has 40 per cent market share in NAND flash and Toshiba has 27 per cent share.

Toshiba is currently ramping a massive fab at Yokkaichi which is expected to be outputting 80,000 wafers a month by the end of the year with the capability of expansion to 200,000 wpm. And, with its partner San Disk, Toshiba is to build another new fab starting in 2009 and scheduled to run wafers in 2010.

So Intel is stuck with a dilemma. Should it pull out of NAND before it has properly got in? Or should it bear the pain of competing with a rampant Toshiba and two other powerful Asian NAND competitors, Samsung and Hynix?

Pulling out will expose Intel to the charge that it can't compete in markets where it doesn't have a monopoly. Staying in will see Intel commit what Wall Street sees as a cardinal sin: eroding margins.