The unthinkable rumours are being all but confirmed. Financial analysts from Needham and Company claim that Nvidia has less than 10,000 cards worldwide at the moment (all sold out, presumably) with yields of 20%-30%. Things could have been even worse, perhaps hitting the “<20%” rumour had GTX 480 shipped with all 512 SP enabled.

In an analysis to their clients, analyst Ravindra Gill wrote – “We are downgrading Nvidia to a “hold” and removing our 12-month price
target (was $22) after a series of channel checks indicating that Fermi
is not ramping well and there could be further product delays. Based on
our findings, Nvidia has very limited supply of Fermi desktop/notebook
parts and yields remain poor at around 20-30%”.

More details…

The unthinkable rumours are being all but confirmed. Financial analysts
from Needham and Company claim that Nvidia has less than 10,000 cards
worldwide at the moment (all sold out, presumably) with yields of
20%-30%. Things could have been even worse, perhaps hitting the
“<20%” rumour had GTX 480 shipped with all 512 SP enabled.

In an analysis to their clients, analyst Ravindra Gill wrote – “We are
downgrading Nvidia to a “hold” and removing our 12-month price
target (was $22) after a series of channel checks indicating that Fermi
is not ramping well and there could be further product delays. Based on
our findings, Nvidia has very limited supply of Fermi desktop/notebook
parts and yields remain poor at around 20-30%”.

Mr. Gill expects delays to continue with delays on derivatives of the Fermi family a definite possibility. Meanwhile, there is little news of Fermi powered Tesla and Quadro. It would make sense that Nvidia would shift focus on these cash cows, and under such exceptionally limited yields, Nvidia might just allocate more GF100/Fermi chips to the more profitable Tesla and Quadro products. Meanwhile, Nvidia is expected to continue to lose market share through the second half of 2010.

We have full faith in Nvidia’s marketing team to sell enough GF100 products, despite their obvious price/performance/watt inferiority, but for that to happen, Nvidia must have some cards for sale. The only real solution for Nvidia at this stage is to do a full base layer respin, as no doubt Fermi’s problems are deep rooted. It does seem that Fermi was “overambitious” and just way too massive for the troubled 40nm process. A die shrink would have been godsent, but unfortunately, the next process 28nm is at least 1 year away.

If Nvidia have fought back from NV30, and ATI from R600, there’s no reason as to why Nvidia can’t get back to the top following the Fermi debacle. If ATI continue progressing, however, Nvidia have steep mountain to climb.

Reference: Xbitlabs