Fermi, GeForce, Quadro, Tesla and Tegra

Nvidia's money man talks about its technology at a recent conference.

At the Goldman Sachs Technology and Internet Conference taking place in San Francisco, Nvidia's CFO, David White, opened up to answer a few choice questions from the host. MCPs are dead for now, Quadros and Teslas are coming, as are lower-grade Geforces, all of which will be based on Fermi GPUs.

According to Mr White, and despite a relatively rough year for the company for lack of new-new products, Nvidia's master plan seems to be falling into place. Supply constraint dictated much of the outcome of the last two quarters, with Nvidia unable to squeeze enough of the necessary wafers out of TSMC to meet demand, much less ramp up production. As AMD had already mentioned, only in December did TSMC get its act together and get with the plan. Supply will be back to normal by mid 2010. But in the meantime everyone in the 40nm chip business will have to bite the bullet.

The company's earnings breakdown also revealed just how money really does make the world go round. Nvidia's biggest cash-cow last quarter, said White, was the Quadro-series of products, which sell at about 10 times the price of GeForce cards but are cut from the same silicon. If you're on allocation and Quadro demand is high, you as a company simply can't afford to build Geforces, now can you? Quadro had the highest revenue, just shy of $200 million, said White.

Technology-wise, the chipset business is flat, thanks to Intel, Nvidia's CFO pointed out. All development activities related to MCP stopped due to Intel's unwillingness to continue chipset licensing and things just went downhill from there. There's nothing on the menu unless Intel unblocks Nvidia. The FTC seems to be Nvidia's best hope right now as it's taken an interest in Intel's business practices of late. You can also infer that Intel can shut Nvidia out of the netbook business very easily. Nvidia has no expectations that things will improve on this front over the course of 2010. "We believe we have a DMI license." added White. "As a result of a lawsuit and other behaviour on their part, [Intel] basically neutered our MCP business and made it difficult for us to achieve the penetration and success in that business that we would've expected from that business."

Going into a bit more Fermi-related news, the moneyman at Nvidia shed some light on plans for Fermi. Simply put, Geforce high-end, Quadro and value Geforce, with Tesla popping up somewhere in between the last two.

We did notice that the Green Goblin's marketing diktat made itself present. Nvidia's CFO dismissed the notion that Fermi is a "compute architecture" just because Jen-Hsun Huang had flexed his muscle at a compute conference. White said that a whole range of products based on Fermi will tape-out (implying that they haven't yet), and will bring Fermi to all segments of Nvidia's portfolio, all the while insisting on the prevalence of ray-tracing and PhysX as key to the firm's "innovation" in graphics. So, Geforce first, then high performance Quadro and Tesla products, and only then in 2010 we will see Fermi-based value Geforces.

Regarding the CPU+GPU solution, Nvidia was fairly critical. White pointed out that both AMD and Intel tried and had to overcome great difficulties, and that their solutions were ones of compromise. Taking a page out of the company's own book, White compared the 3 billion or so transistors in the GF100 to the task of developing something like a GPU. If it's hard to do a GPU, it's more than doubly hard to do both in one, and you can never get around the need to compromise. Lastly, Nvidia's White said the firm wants to "Innovate and not integrate." Integration is anti-upgrade, it's a "compromise world by definition." "We don't think it's time to integrate," he said.

Last up, Tegra 2 and its apparent lack of charisma with OEMs. The Green Goblin has plans for Tegra 2. It's the Swiss army knife that other chip makers can't compete with. TI, Motorola, and others are all stuck behind an application processor, but Nvidia thinks this is not the way to go. It's looking at what people would like to do with their devices, very much like Apple did when it developed the Iphone. Smartphones will be key to the success of Tegra.

However, the biggest design coming out of Nvidia's labs soon is a Tegra 2-based tablet. In fact, Nvidia believes the Tegra 2 will challenge Apple directly.

So cutting a long story short, Nvidia's chipsets are dead in the water until the FTC comes to the rescue, Fermi is firmly locked-in as the Swiss army knife for everything at Nvidia except the Tegra 2 that's meant to challenge the Iphone.