CNET también está disponible en español.

Ir a español

Don't show this again

Tech Industry

2HRS2GO: 3dfx buried itself

    COMMENTARY -- 3dfx Interactive (Nasdaq: TDFX), we hardly knew ye.

    It's tempting to use that in a eulogy for the 3D graphics pioneer, except that people know it all too well, and they don't like what they see.

    3dfx on Friday announced the last stage in a slide that began more than two years ago. The board decided to sell the guts of the company to current 3D king Nvidia (Nasdaq: NVDA), then take up permanent residency in the tech company graveyard.

    No one can say what 3dfx investors get out of this deal, because the company plans to take care of its debt, which totaled more than $118 million at the end of October. Don't put your hopes in getting NVDA stock -- the 1 million shares offered by Nvidia would be considered part of 3dfx asset liquidation prior to settling with creditors.

    On the other hand, anyone who has remained a 3dfx shareholder since last year has irrational hopes anyway. 3dfx suffered long before the downturn in the PC market; lousy quarterly reports have been a 3dfx staple for six of the last seven quarters.

    "Long since two months-post IPO and holding on 'til they pry my cold, dead fingers from my shares," was the self-description of one poster on Yahoo! Finance's 3dfx message board.

    You can't reason with that kind of person. Fortunately, most of us -- including one columnist who wrote a stupidly optimistic piece on 3dfx back in May 1999 -- can look more rationally at 3dfx.

    Market research analyst Jon Peddie notes that it's easy to point fingers now -- "This is Monday morning quarterbacking, and it's not a totally fair thing to do" -- but 3dfx still deserves criticism. "They didn't have the right combination of manufacturing and engineering in time," says Peddie, whose firm specializes in graphics and multimedia.

    Many of 3dfx's problems were self-inflicted. Not that unintentional hara-kiri is anything new in the graphics controller industry, which at its peak had as many as 45 companies, by Peddie's count.

    "Every company, every single one of them, that has dropped out or failed, has been done in by its own hand," he says.

    Peddie believes 3dfx was the victim of its manufacturing ineptitude. The company jumped into the graphics card vendor business by agreeing to buy STB Systems in 1998, right about the time when PC vendors began clamoring for a card that could handle both 2D and 3D functions. Until that time, most computers had separate cards for 2D and 3D.

    But 3dfx didn't have a 2D/3D chipset available immediately, so original equipment manufacturers turned away from STB and went with cards featuring technology from Nvidia or ATI Technologies (Nasdaq: ATYT).

    "3dfx didn't have anything that the OEMs would want," says Mike Feibus of Mercury Research. "It's always been that you're only as good as your last chip. 3dfx fell faster becasue they didn't have OEM design wins to ride in the main PC market."

    By the time 3dfx shipped Voodoo 3, Nvidia and ATI were firmly established in the PC market. 3dfx then componded its mistake by shipping the next Voodoo generation seven months late.

    Not only did 3dfx miss the holiday season entirely, but when the Voodoo 5 came out, it looked overpriced for a card whose raw performance trailed that of Nvidia's GeForce2 and ATI's Radeon products. 3dfx tried to sell users and OEMs on the Voodoo 5's other features -- mainly "full-screen anti-aliasing," which makes a 3D scene look smoother -- but reviews were mixed at best.

    "Although we strongly believe that visual quality will be the benchmark for future success, we did not sufficiently establish visual quality as the most important attribute for consumers to rely on when making their purchase decision," 3dfx CEO Alex Leupp admitted during Friday's conference call with analysts. "Therefore, our competitiveness was impacted, which resulted in lower-than-expected revenues, as well as pricing pressures that were not anticipated."

    Even then, 3dfx might have survived long enough to borrow more money, had PC sales not slowed. But PC demand did slump, and that was the final blow for the already-reeling 3dfx.

    Though Peddie blames 3dfx for its own downfall, Feibus argues the real problem was improved competition. "I don't know that I would put so much blame on 3dfx," Feibus says. "Nvidia set a new bar."

    Since coming out with its TNT chips in 1998, Nvidia has stuck to a schedule of new technology introductions every six months. "Over half of Nvidia's success is due to its manufacturing," Peddie says. "You've got to have great on-time manufacturing. 3dfx never had that. ... That's just discipline."

    Shares of Nvidia are up more than 5 percent today on the latest news. But you have to wonder if there's much to be optimistic about.

    Nvidia eliminated 3dfx as a significant competitor a long time ago, so Nvidia's market share isn't going to get much of a boost from this. And on a broader note, you could reasonably ask if the market for 3D accelerators hasn't peaked, from a Wall Street point of view.

    ATI and Nvidia are branching into other fields. ATI is the leader in 3D graphics for laptops and landed the contract for Nintendo's Dolphin game console. Nvidia is also moving into mobile graphics, and won the accelerator spot for Microsoft's X-Box platform.

    Incidentally, that Nvidia win helped speed 3dfx's death. "3dfx bought Gigapixel, a company with really good technology, but they bought it for the X-Box," Feibus notes.

    No one knows if 3D companies' diversification efforts -- Nvidia and ATI are also looking at system logic and other areas -- will turn out to be as lucrative as the PC market has been. They'd better be rich fields, because the 3D graphics business on desktop PCs isn't growing like it was three years ago, Feibus points out.

    "The pot of gold at the end of the rainbow has declined," he says. "I think investors know it."

    Perhaps Wall Street has factored that in already. Nvidia currently trades at 23 times estimated earnings for the next 12 months, or comparable to Intel (Nasdaq: INTC) at a multiple of 22.

    The near-term success of Nvidia and ATI will depend on how well they can raise the price of their products, Feibus says. They've been successful so far; the average price of the highest performing 3D chips available have doubled over the last 18 months, according to Mercury Research. Nvidia's GeForce 2 Ultra sells for around $400 or higher.

    Peddie scoffs at any notion of a brick wall for the growth of PCs and related fields. He sees nothing wrong with a "maturing business with reasonable margins." 22GO>