Nvidia looks good

Posted: 01.11.11

Nvidia like many companies in the past two years has gone through some dramatic and bruising changes. How a company comes out of those events is the key to predicting their future.

I’m not going to review the past, if you don’t know it, you probably aren't reading this.

Here’s what I see on the horizon for Nvidia.

Tegra, the company’s smallest product, smallest ASP, smallest results to date is probably going to show some life in 2011. The company is shipping Tegra 2 and will soon introduce Tegra 3 to OEMs – the rest of the world will not get much information on it anymore than they did on Tegra 2. News and leaks of design wins in phones and tablets have been dribbling through webosphere for the past nine months as the industry waited and watched for the results Nvidia predicted at their press conference at CES in January 2010. Few of those predictions happened and a lot of people wrote off Nvidia as a player in the SoC market. But one of the qualities of a great company is that they don’t quit; they keep pushing and redoing till they get it right. Nvidia believes in Tegra and has not only kept up its investment in R&D and marketing, but has actually increased the spend. That’s commitment and it’s got to pay off – 2011 may show the results.

News of Microsoft making Windows run on ARM will help propel Nvidia’s fortunes in the SoC space.

Tesla and CUDA. 2010 was actually the tipping point for Tesla. For evidence consider the almost clean sweep of HPC companies with Tesla based server chassis, starting with IBM, HP, Penguin and followed by others. Vector or parallel processor competitive products either failed, or flailed. Cell, FPGAs, direct GPU competitors, dedicated multiprocessor processor arrays, and multicore CPUs couldn’t deliver the bang for the buck in FLOPS/dollar/watt, nor did they have the infrastructure and tool set to make them useful. Nvidia’s long term and unwavering investment in Tesla and its CUDA tools are paying dividends. Although the HPC element will not be a big unit or dollar market it will be significant, measurable, and most of all influential. GPU-compute will not be limited to HPC and will find homes in everything from photo-editing to plastic molding. GPU-compute will also be a very high margin segment for Nvidia, and that contribution makes all the products better.

Fermi. Fermi is finally firm. After the redesign in 2010 which yielded the 500 series parts, Nvidia has a world class GPU again, and one with legs that will let it find product categories from laptops to mainstream PC, to gaming boards, to professional graphics and HPC. Fermi, like Tegra, was a long term, and difficult investment, with a very long range vision behind it. The company suffered severe criticism for abandoning the mainstream and gamer with the original 400 series Fermi design, and for being late with their DirectX 11 part. AMD capitalized on that and increased their market share. There was speculation that Nvidia had been scared out of the mainstream by Intel’s Larrabee and was concentrating on different areas. Chip design, and re-design doesn’t happen overnight, and so Nvidia’s redesign of Fermi could have been a reaction to the failure of Larrabee to arrive. Regardless, the 500 is a solid product and Nvidia will exploit the design for several years.

Quadro the cash cow. Nvidia has historically done very well financially in two small highly profitable segments: the enthusiast gamer and the professional graphics markets. At its peak Nvidia had 65% of the enthusiast gamer segment (now at about 55%) and has gone from 20% in 2004 to 92% in 2010 in the professional market. The growth was the result of Nvidia's investment in the segment, support for the application developers, and paying attention to users. Nvidia got a little help from ATI’s difficulty in focusing in the workstation space. We expect Nvidia to hang on to a dominate position in professional graphics, but with such a tremendous lead in market share they will logically lose a few points. Nvidia gets better margins for their professional graphics than AMD.

Stereovision and Optimus. Nvidia was the leader in stereovision beginning in 2009 and still is, however, it’s just a small market (~300k units). But as stereovision grows, and it will, Nvidia’s brand is so strongly identified with it, that most people think Nvidia invented PC stereovision. Meanwhile, Optimus is Nvidia’s ticket to regaining its lost laptop market share. Optimus is a truly useful feature for anyone who needs high performance and long battery life and it has been adopted by several OEMs, along with Intel’s new Sandy Bridge EPG CPU.

Intel settled. The recent settlement with Intel that will put $1.5 billion into Nvidia’s bottom line is just one more bonus for the company removing management distraction and padding the profits. It’s also a moral win for the company.


Nvidia has invested heavily over the past two to three years in the company’s vision, a vision of “T”s – Tegra and Tesla. It suffered after being shut out of the IGP market by Intel’s claim of limited license, and it lost significant market share in laptops with its discrete GPUs as well. During those difficult years the company, reported its first losses in a decade; its share price plummeted and pundits described the company as being on the ropes. Maybe, but it survived (the second time), didn’t have massive layoffs, and is poised to enjoy the improving economy with a product portfolio that is very impressive.