PALO ALTO -- Don't buy chips yet.

Dueling chip analysts Dan Niles (

Lehman Brothers

), Mark Edelstone (

Morgan Stanley

) and Drew Peck (

SG Cowen

) didn't lend a lot of positive vibrations to the eager crowd at the sixth annual semiconductor forecast sponsored by the Churchill Club, a Silicon Valley public-affairs nonprofit. For those of you looking for hope, you won't find it here. The trio was in comfortable agreement that the worst is not over.

At last year's forecast the three were pressured to call the top of the semiconductor cycle. This year the heat was on to tell investors when to expect galloping gains again. No such luck: The analysts agreed that those up, up and away chip stock charts of the '90s should be framed, because they'll give investors misty-eyed memories to enjoy when the volatile, but flatter "trading range" movement of the 1980s chip industry returns.

Looking ahead at the next few months, instead of the whole decade, Niles said we're in for a "U"-shaped recovery and a deeper bottom than most are expecting. Edelstone gave an example of that by noting that manufacturing plant utilization rates that were up to 96% last year are below 60% now and won't even hit 80% until the second half of 2002. With manufacturers twiddling their thumbs and inventories wrecking balance sheets across the sector, Peck advised that the only way to invest right now is to pick stocks that still have end consumers -- stay away from chipmakers with wares that go into cell phones or PCs.

Edelstone assured the crowd that the chip business will come back in the third or fourth quarter. However, he cautioned against expecting the fourth-quarter's holiday season to make everything all right, because he expects all the cell phones and PCs bought at Christmas will be built with stockpiled inventory. Edelstone estimates that chip inventories are still running at 100 days, the highest levels since 1989. All this means, as Edelstone explained, is that "margins won't improve for multiple quarters." Low margins mean companies will have trouble making their earnings -- as 80% did in the first quarter, according to Edelstone. He expects the stock prices to stay down until investors see positive earnings surprises.

Peck relied on charts to put forth his trademark dour predictions. He explained that semiconductor busts come with two peaks. The first and highest is what we saw in late 2000 and the second is coming now. He classified the most recent uptick in chip stocks as the snapback that occurs when inventory and order levels dive far enough below chip consumption. Once things normalize again, we'll be headed downhill for a while, as Peck illustrated by showing a chart of the 1995 and 1996 double-peaks.

"People who bought in at the end of 1995 got their heads blown off," Peck stressed. Take home the point: Those hell-bent on early re-entry into chip stocks got crispy and creamed.

Showing the form that made him winner of the event the previous five times, Niles explained that just because a stock may have fallen 50% it is not by definition cheap. Using

Intel

(INTC) - Get Report

and this week's prime example,

Sun

(SUNW) - Get Report

, he warned that it can be more painful to jump back into a stock before another fall-off than to mistime the exact moment it starts climbing in recovery. He expects Intel to warn again in the future, and said that while Intel has fallen dramatically, by historical standards it's not cheap.

Be patient, be picky and don't expect any miracles.

Which brings us to the blood sport of the evening, the analysts' weird and wonderful stock picks. Investors should take these with a salt-shaker in hand given that in last year's analyst-eating market the best long pick among the three was

Micron

(MU) - Get Report

, which fell 47% and was picked by now five-times winner Dan Niles. His

NETsilicon

(NSIL)

short pick saved him by dropping 46%, but it's more than obvious from a quick review of 2000's picks that while all three analysts have plenty of experience witnessing semiconductor cycles, they're a lot easier to dissect when they're over. It's also worth noting that the three kept to the familiar: Each analyst picked a company for which his firm did investment banking.

Kicking off a new year, Peck was mock-strangled by Niles when he selected

Lucent

(LU)

spinoff

Agere

as his long stock because of its size and potential. Niles used his backup,

ON Semiconductor

(ONNN)

, for a long, while Edelstone showed fidelity, if not a tolerance for pain, by picking

Broadcom

(BRCM)

to go up in the next year.

The analysts were sure to note that this was a 12-month contest, and while they all expect their stocks to have trouble in the next few months, over the 12-month time frame they expect them to thrive.