In part one of our investigation into the current health of Nvidia, we painted a pretty downbeat picture.
Nvidia's desktop and laptop chipset businesses are being pinched, Ion looks like a dead duck even as it's being launched, Nvidia's core PC graphics business is in the trenches, Tegra is a virtual non-starter and its console cash flow will die with the underperforming PlayStation 3.
But things are often not what they seem in the computer chip industry. Lest you have forgotten, 12 months ago AMD looked utterly dead in the water. Today it has arguably the best graphics chips in the world and a much more competitive CPU product in the Phenom II.
Where better, then, to get the case for the defence than Nvidia itself in the shape of Derek Perez, the company's long serving Director of PR? If Perez can't spin Nvidia's current situation into a tale of impending glory, nobody can.
The first thing Perez emphasises is that the PC industry is on the cusp of a sea change that will play directly to Nvidia's strengths.
"Applications are now appearing that really begin to leverage the parallel computing of graphics chips. Even if you take away conventional graphics processing and gaming, parallel processing is becoming the norm. And big new markets like HPC (high performance computing) in industries such as oil and gas are just starting out," Perez reckons.
Unsurprisingly, Perez says Nvidia is best positioned to cash in on this industry-wide shift towards parallel computing, or as some would have it, visual computing: "The bet about the shift towards general purpose computing on the GPU is not if, but when."
No evidence to support chipset threat
As for the threat to Nvidia's chipset business from CPU-GPU fusion chips, Perez is totally unconvinced. "There's an assumption in all of this that Intel's on-CPU graphics will be good enough. But there's absolutely no evidence to support that view," he says.
And he has a point. The history of Intel's integrated graphics cores is one of mediocrity bordering on awfulness. But isn't it hard to imagine a PC that has two integrated graphics cores - one in the CPU itself and one on the motherboard? "If we continue to innovate with great products like the GeForce 9400M, our chipset business will continue to thrive," says Perez.
But what of Ion? Surely the arrival of Intel's upcoming Moorestown system-on-a-chip replacement for the current Atom processor will kill that chipset stone dead? "Just because Intel will build it, that doesn't mean it will perform as Intel claims or that it will be what people want. How good will the graphics and audio be in Moorestown?" Perez asks.
Moreover, Perez says 40 per cent of netbook returns are from unsatisfied customers and much of that is due to poor video performance. What people need is not a better Atom processor, but a better graphics core, which is precisely what Ion delivers.
That may be true for netbooks. But Ion is not a viable alternative to Moorestown for truly compact devices. It's simply too power hungry. Which brings us neatly to Tegra, Nvidia's chip for smartphones. Perez admits Nvidia's inexperience in the mobile phone market may have led the company to underestimate the time it would take to win over the market with Tegra.
"Our PC mentality told us to think in terms of a year," he says, "but the reality for mobile phones is more like five years." Still, Perez claims we will all be able to buy a MID or phone powered by Tegra no later than Christmas this year. More importantly, he is adamant that Nvidia will announce a partnership with at least one of the big mobile phone makers within a year.
The core PC graphics business
As for Nvidia's core PC graphics business, Perez agrees that the momentum is in the $120 to $199 range and that its entry-level discrete business is increasingly under threat, not least from Nvidia's own GeForce 9400 integrated graphics. But he also thinks that there's plenty of potential for the GeForce 9400M integrated to cash in on the burgeoning $399 to $599 notebook market.
In the next six to 12 months, that is undoubtedly true. But beyond that, it's hard to see how the introduction of Intel and AMD's CPU-GPU chips won't do serious damage to the 9400M's prospects, no matter how awful Intel's integrated graphics may turn out to be.
We are not, therefore, entirely convinced by Perez's pitch. Nvidia's chipset business will definitely be pinched by the coming era of fusion processors, the only question is by how much. And that inevitably leads to the greatest Nvidia-related question of all. Will the company eventually be forced to somehow get in on the x86 game and produce a CPU core of its own?
Here, Nvidia is currently putting out mixed messages. Perez dismisses the idea, saying, "We didn't need x86 for our first 15 years and we won't need it for our next 15." However, Nvidia VP Mike Hara recently told an audience of investors and analysts that it would eventually make sense for Nvidia to do with x86 what it has done with an ARM core in its Tegra mobile phone processor.
In other words, include a general purpose x86 core within a larger system-on-a-chip design. The only problem with that idea is that of the x86 core itself. Where would it come from? While ARM is happy to license one of its cores to Nvidia, it's hard to imagine AMD or Intel doing the same. Which leaves the likes of Via as a possible, but distinctly underwhelming, potential CPU core supplier.
Well, that's assuming Nvidia is unable to pull off a spectacular legal coup and produce an x86 core of its own without incurring the litigious wrath of both AMD and Intel. Whatever happens, the next few years will be literally make or break for Nvidia.
Like this article? Then check out The 10 best graphics cards money can buy
Sign up for the free weekly TechRadar newsletter
Get tech news delivered straight to your inbox. Register for the free TechRadar newsletter and stay on top of the week's biggest stories and product releases. Sign up at http://www.techradar.com/register