For several years, Nvidia CEO Jen-Hsun Huang has been on a mission to transform the specialized graphics chip into a general purpose “co-processor” routinely installed in the millions of personal computers sold every year.
This year he is more strident than ever.
Costs are higher, but company making a fundamental contribution to computing, says Jen-Hsun Huang
Huang’s belief is that computing itself will be transformed. Scientific and other computationally heavy programs will run faster. New complex applications, such as facial recognition, will be anbled.
“I absolutely believe people will stop thinking about GPUs as graphics processors in the long term,” he told analysts during a meeting Tuesday at the company’s Santa Clara headquarters. “We want to make a fundamental contribution to the computing industry.”
Huang argues that graphics chip maker Nvidia is seeing progress toward its vision. It claims 42 design wins for its new Tegra chip – a product designed for phones and portable Internet devices – with some Tegra smarphones expected in the market by the end of the year.
Nvidia calculates its opportunities will dramatically expand as a result – to an addressable market for all its products of $38 billion compared with $5 billion today.
But the initiative is a gamble, partly because of higher demands for research and development. “Our cost is slightly higher,” Huang admits. This is in part due to $1 billion in excess inventory Nvidia accumulated during the global downturn. “When the economy froze, we were caught kind of flat footed” and product inventories swelled, he said.
As the “co-processor” strategy plays out, “we believe earnings will come (and) margins will come,” he says.
Wall Street analysts have been skeptical so far because of Nvidia’s higher costs and because competitor AMD appears to be rushing new DirectX 11 graphics products to the market ahead of Nvidia.
But Huang isn’t ready to reverse course. “I really believe in what we are doing,” he says And besides, “We are seeing real traction” in the market, he noted, at a time when the next generation operatings systems from Apple and Microsoft will for the first time contain support for “co-processing” computing.
