On February 15th Nvidia published their results for FY2012, ended 31th January 2012 and the results were a mixed bag. The company continues to make a lot of money, which is a good thing, but there are some clouds on the horizon.
The first cloud is that GPU revenues for the year only grew 0.6% compared to FY11. The caveat to GPU revenues in FY12 is that they have a welcome addition of $66 million per quarter courtesy of the Intel settlement, which means that the GPU revenues are boosted by $264 million per year ($220 million in the first year), free of any costs. Exclude that and you are left with a 8.1% decline in GPU related income. This means that pure GPU revenues are falling for the third consecutive year.
The next cloud is Tegra, it did not perform very well as sales in Q4/12 declined 42.5% QoQ. We don’t expect Tegra 3 sales to pick up as A15/Krait offers both huge improvements in performance and offers power/performance than Tegra 3. If you really want top CPU performance you will go with Intel’s Atom.
Last year Nvidia published a very interesting tidbit of information in their 10K filing. While the bulk of their revenue comes from GPU sales, they account for a mere 30 million in operating profits. Compare this with the professional segment, comprising Quadro and Tesla, which had sales of $818 million and accounted for $321 million of the company’s operating profits. As pure GPU revenues fell this year, global gross margins, if calculated to exclude Intel payments, fell to 49.1%. We expect operating results from their GPU segment to stay under pressure for the foreseeable future. As volumes fall, there will be increased pressure on their most lucrative segment, PSB (Quadro and Tesla).
PSB has their silicon developed and paid for by the GPU division, so their major cost of doing business is free as long as the GPU business can afford to support it. Nvidia is about to field a new generation of GPU chips starting in Q2. If they can establish a reasonable performance margin over ATI, and avoid the availability issues that plagued the first Fermi, then they might increase operating margins. Unfortunately, sales are likely to drop as Intel and AMD field a new generation of more powerful IGPs and the Ultrathin format gains traction. Aside from the occasional stunt, the ultrathin form factor will not support a discrete card from Nvidia, and the GPU bearing CPUs (IGP) will eat into the GPU business in ever increasing amounts.
The Tegra product line booked an operating loss of 150 million for the entire FY12, and had its sales outlook for FY13 reduced from $1 billion to $540 million. If they can achieve this projection, it may be enough to break even given the ever rising capex needed to support those sales. If Tegra 3 fails to gain sufficient traction as is currently looking likely, that breakeven point will be very hard to reach in FY13.
Nvidia is correctly focusing most of their efforts in their Professional Solution Business segment. As with Fermi, their top chip will offer compelling graphics performance, but it is on GPU compute tasks that the chip shines. The new generation, Kepler, is said to maintain this level of compute performance , but is not due until much later in calendar 2012. While the workstation line is fairly secure as Intel has not produced a product for this market and AMD has yet to deliver a competitive solution, the HPC market will be attacked by Intel and its Knights Corner chip. We expect margins to hold steady here as Tesla represents just a small part of the PSB.
Clouds continue to hang over Nvidia’s future as they transition product mixes and business focus, whether or not their new strategy yields success remains to be seen.S|A