Nvidia tries to snow financial analysts

31 Jan 2008 | 09:35 GMT

By Charlie Demerjian

SO, HOW LOW can Nvidia go? Well, how about official messages from its Vice President of Investor Relations, Michael Hara that sink below the low, low standard the aggressive outfit normally sets.

Luckily, unlike many other episodes, we can't find an outright lie in this set, but then again, the audience is a little more touchy about that sort of thing.

The letter, printed in full below, was sent off by Mr Hara Wednesday at almost 1pm PST according to the timestamps, and it is a doozy. In addition to this, we have seen a 22 page PDF entitled "3870 X2 Launch Analysis", which most people would call a one sided-joke.

That said, first the letter, with headers intact so you can see that 1) Mr Hara is an idiot for posting his mailing list publicly. 2) You can see who he mailed. 3) You can reply to him as well.

Here it is, we removed the .coms from mailing addresses so they would not be automatically harvested, but it is otherwise in tact save for minor readability edits like extra white space removal.

---------------------------------------------------------------

From: Michael W Hara <MHara@nvidia>

To: Amanda Hindlian <amanda.hindlian@gs>; Andrew Buscaglia < abuscaglia@thinkequity>; Arnab Chanda <arnab.chanda@db>; Bobby Gujavarty <bobby.gujavarty@db>; Brian Piccioni <brian.piccioni@bmo> ; Craig Berger <cberger@fbr>; Daniel Ernst <daniel@hsqr>; Daniel Gelbtuch <daniel.gelbtuch@us.cibc>; Daniel Morris < daniel.morris@us.cibc>; David Hodgson <david.hodgson@genuitycm>; David Wu <david.wu@globalcrowncapital>; Devan Moodley < devan_moodley@scotiacapital>; Dinesh Moorjani <dmoorjani@bear>; Doug Freedman <dfreedman@amtechresearch>; Elizabeth Pate <epate@fbr>; Garik Shmois <gshmois@longbowresearch>; Glen Yeung < glen.yeung@citigroup>; Gordon Johnson <gordon.johnson@lehman>; Gurinder Kalra <gkalra@bear>; Hans Mosesmann <hans.mosesmann@raymondjames>; Heidi Poon <hpoon@tweisel>; James Covello <james.covello@gs>; Jan-Jesse Flores <janjesse.flores@lehman>; Jason Pflaum < jpflaum@tweisel>; Jean Orr <jorr@nutmegsec>; Jean-Pierre Becker < jean-pierre.becker@tdsecurities>; Jesse Tortora <jesse_tortora@prusec> ; JiHyung Yoo <jihyung.yoo@citigroup>; Jim Karlis < jkarlis@williamblair>; Feeney, JoAnne <joanne.feeney@ftnmidwest>; Joel Jackson <joel.jackson@bmo>; Kamal Das <kamal.das@lehman>; Krishna Shankar <kshankar@jmpsecurities>; Larry Cao <larry.cao@morningstar> ; Michael Masdea <michael.masdea@credit-suisse>; Michael McConnell < mmcconnell@pacific-crest>; Naser Iqbal <niqbal@salmanpartners>; Nicholas Aberle <naberle@cariscompany>; Peter Karazeris < peter.karazeris@citigroup>; Quinn Bolton <qbolton@needhamco>; Rick Schafer <richard.schafer@us.cibc>; Robert Burleson < rburleson@thinkequity>; Ryan Elliott <ryan.elliott@tdsecurities>; Shawn Webster <shawn.r.webster@jpmorgan>; Sidney Ho <sidney_ho@ml>; Simona Jankowski <simona.jankowski@gs>; Suji De Silva <sdesilva@kbro >; Tayyib Shah <tshah@longbowresearch>; Terry Thib < terry.thib@nbfinancial>; Thomas Smith <thomas_smith@sandp>; Tim Luke <tim.luke@lehman>; Uche Orji <uche.orji@ubs>; Will Gabrielski < wgabrielski@amtechresearch>; Yuka Miyake <yuka.miyake@gs>

Sent: Wed Jan 30 15:53:29 2008

Subject: Analysis of Radeon HD 3870 X2

On Monday, January 28th, ATI launched the Radeon HD 3870 X2 graphics card. This is a dual-RV670 board that ATI is positioning at ~$449-499. Attached is a comparison presentation that outlines the advantages of the GeForce 8800 GPUs over the Radeon HD 3870 X2. There are a couple important things to keep in mind regarding the 3870 X2:

1) Radeon HD 3870 X2 relies on CrossFire driver technology, and thus has the same poor driver support that CrossFire has suffered from including:

- Poor support for next gen DirectX 10 games at launch

- Poor scaling and compatibility on games that are not top benchmarks

2) ATI has not been able to get their Quad drivers for 3870 X2 working and has delayed driver support for QuadFire until March(!). This indicates they are having serious problems with Quad support. GeForce 8800 in SLI remains the undisputed fastest gaming platform on the market.

3) GeForce 8800 GTX prices have been coming down, with overclocked versions readily available for <$449. As you can see from the attached presentation, GeForce 8800 GTX delivers superior DirectX 10 gaming performance than Radeon 3870 X2, and with overclocked 8800 GTX's available for less than Radeon 3870 X2, 8800 GTX continues to deliver the best price/performance. Even the GeForce 8800 GTS 512MB beats Radeon HD 3870 X2 in a number of games and at ~$299 delivers outstanding price/performance.

Enjoy. <<3870 X2 Launch Analysis.Jan08.pdf>>

Michael Hara - NVIDIA Corporation - Vice President of Investor Relations - (408) 486-2511 - Fax: (408) 486-4511

--------------------------------------------------------------

Lets parse this for a minute. First he artificially ups the price, calling it from $449-$499. For the past few days, all six of the cards listed at Newegg are for sale at $449. There isn't even one available for $450, much less $499.

If you want to try a futile exercise, may we suggest you look at the same site and try to find a single G92 card for their supposed retail price point, start here. Now go back and look at the slides they gave out at the launch. See how they are using you? Remember this next time they promise you something on a call.

On to Point 1, that the X2 has the same 'problems' as Crossfire. True. So does normal Crossfire, and Nvidia's much-pimped SLI. Same problems, it is a game engine scaling issue, not a card issue. NV's upcoming dual cards will suffer just as badly, likely worse due to inferior construction and layout.

That said, ATI has cards out now, and NV could not show working boards at CES, product launches were not touted to the press at CES editors' day either. When you don't have what it takes, spin to the ones that matter. If I was an analyst with a clue, I would be annoyed at such blatant, at best, ill-intentioned cobblers from Mr Hara.

Next, poor scaling on DX10 games, evidenced up by the slides NV sent out. We won't bore you with the details, but strongly suggest you go read the FULL articles they quote. We have not seen a more disingenuous case of cherrypicking since, well, the last set of NV lie^h^h^hslides.

For example, the second slide, the one after the title screen, links to a single slide on this page. The quote cherry picked is: "It's not flattering for the HD 3870 X2, which is behind its two competitors from NVIDIA. An inefficient Crossfire appears most probably at fault, the increase compared to a 3870 being limited to 23% at best". Sounds bad, right?

You will notice that the vanilla 3870, a $250 card, spanks the Geforce 8800Ultra, a nearly $700 card, in several areas. Pick the one you want, but it is far from what NV claims, and its conclusion of better at DX10 is not even close. Then again, NV seems to have picked the one slide that backs its argument from a 20-page review with at times multiple slides per page. Funny that.

What does Toms say in conclusion? "The fastest card today, once again, carries the AMD signature, after a long domination by NVIDIA. Based on two RV670, the HD 3870 X2 suffers, however, in part from its conception and isn't as thrilling as a new chip, like a real R680, would have been." You can read it all here, no cherry picking needed.

Point 2 is the quadfire drivers. Yup, they are delayed, but they are not 'not working', I have seen them in action a few times personally, and I am guessing it is a polish and tuning thing. There have been several public demos, most notably at the NYC analysts' day that most of the CC'd analysts above attended. Then again, NV has never so much as shown me working quadSLI, and it has been promising that for years now. Place your bets on who gets there first, but "not working" is a rather callous half-truth at best.

Point 3 comes back to the same cherry picking and price sleaze as Point 1. If you go to Newegg, there are 22 hits on the 8800GTX, including open boxes. Of these, two are under the quoted price, two are at it, and the rest are substantially above it. Remember every 3870X2 is at the lowest point of the 'range' Mikeypoo quoted, and if you hold him to the < (IE less than, not <=), 2 of 22 hit his price point. None of these is overclocked like he claims, those versions start at far higher prices.

Can you say credibility gap? I would say he missed a page in his research, but any fool with half a brain can verify his claims, or as usual, shoot them down, in less than five minutes with a browser. He also doesn't mention that the GTX he is so hot on was EOLed this month. And the ~$299 8800GTS he crows about, well Newegg has 14 flavours of it, only one of which hits the price point he claims, and the rest are way above that.

We won't go through the details of it being the fastest, that one is so laughable it isn't worth bothering with. Go read all of the reviews he is quoting, each page, all the way to the end, and then make up your own mind. If you are really bored, go find where he cherry picks and what was said around that.

Then ask yourself this, if they are willing to go to such lengths to bend the news to the financial analysts through official channels, what else are they bending? If you are in the mood to make Mr. Hara squirm, ask him about NVs relationship with Intel right about now, and then ask Intel the same questions. How much of their business is based on Intel chipsets?

Ask him about OEM GPU design wins for Q4-Q2. Ask him about Montevina notebook wins. Ask ATI the same questions. Ask NV about its CSI licence status. Ask why it are more than six months late delivering a real PCIe2 product that fully works. Ask why it is having such trouble geting chipsets working with Intel 45nm parts. Ask about.... well, that should be enough for now, why spoil all the fun?

When you are done with this, run the numbers with their numbers, and run them with Intel's and ATI's. See any difference? You are analysts, right? Then ask yourself which side has credibility. µ

© 2007 Incisive Media Investments Ltd. 2007