Five years ago, NVIDIA (Nasdaq: NVDA) was on top of the world. The GeForce series of graphics chips was smashing the competition on important metrics like performance, bang for the buck, and sales. NVIDIA had a healthy secondary market in chipsets for Intel (Nasdaq: INTC) and Advanced Micro Devices (NYSE: AMD) processors. Looking into a crystal ball, AMD was about to buy out main rival ATI in a messy acquisition, with years to go before any real payoff. Everything was looking up.
Dark clouds
That was then; this is now. The wheels started falling off NVIDIA in early 2008, as production problems with the new G92 graphics platform led to slow sales and heavy write-offs. At the same time, Dirk Meyer took the reins of AMD from perennial underperformer Hector Ruiz, and started whipping that sorry horse into shape on many levels. ATI now rules the roost from value-priced starter cards to many high-end enthusiast products, and NVIDIA's production problems have resurfaced time and time again.
Oh, but that's not all. The chipset business is not much of a contributor to NVIDIA's business anymore, because a public spat with Intel over licensing issues led to angry words and lawsuits; the company is no longer allowed to design new Intel-compatible chipsets; barring new legal arrangements, the company will see that segment dry up in the coming years.
And it's easy to see this series of events playing out by looking at NVIDIA's results. Trailing-12-month sales topped out at $4.4 billion in the first quarter of fiscal 2009, before dwindling to $2.8 billion five quarters later. By then, NVIDIA was unprofitable, and the shares had lost about 75% of their peak value from early 2008. Garbage-diving investors could grab shares for as little as $6 to $8 a stub, hoping for a recovery.
Blue skies? Nope, still stormy!
NVIDIA began its recovery fueled by high hopes around the Fermi architecture, Tegra-branded mobile processors based on technology from ARM Holdings (Nasdaq: ARMH), and the Tesla supercomputing product. Microsoft (Nasdaq: MSFT) launched a Zune HD media player built on Tegra chips, and the clouds didn't seem so dark anymore.
Alas, there always seems to be a "but" when it comes to NVIDIA. This time, Fermi chips turned out to be difficult to manufacture, and hard to run without cooking eggs on your graphics card. Texas Instruments and Qualcomm (Nasdaq: QCOM) are battling it out for supremacy in the ARM-processor space, with NVIDIA nowhere in sight, and even the Zune was a flop.
The company's recovery stalled at about $18 per share, and a severe revenue-growth slowdown around $3.7 billion on a trailing basis, and now the stock is back to less than $10 a share.
What can NVIDIA do?
The cure for NVIDIA's many ailments seems simple: Stop the "overpromise, underdeliver" game and start doing the opposite. CEO Jen-Hsun Huang talks a big game in good times and in bad, but only seems to deliver when he can lean back on heavy tailwinds. That's not how you become great, or how you stay on top once you get there.
If I were running the show at NVIDIA today, I'd take a step back from the hype-o-mania that always swirls around this company. Let the stock sink until NVIDIA gets its sea legs back.
The balance sheet is mighty strong after many good years and only a few lean ones; NVIDIA has $1.8 billion of debt-free cash in the bank. Invest a good chunk of that in R&D efforts in core graphics areas, hiring proven engineers and top graduates away from the competition. Put most of them to work on process technology in conjunction with a support team from main manufacturing partner Taiwan Semiconductor Manufacturing (NYSE: TSM), with the rest boosting the next generation of graphics chips. Don't change this model until NVIDIA again beats ATI and Intel on every important metric, from cost-effectiveness to power efficiency. Most importantly, manufacturing yields on these chips should be excellent from Day One.
That's when investors get the payoff from this difficult transition, especially those who picked up shares on the cheap.
That's a far cry from how NVIDIA operates today, but I think it can be done. Would it be easy? Heck no! It'd be a tooth-and-nails battle in a steel cage with no holds barred, fought while AMD is trying to rewrite the rules to the whole game. Is it worth a try? Absolutely.
Will Huang do it? His operating history and larger-than-life persona tells me there's no chance at all, and that his company is doomed to eternal mediocrity. Let me know in the comments box below whether you agree -- or whether you think I should give NVIDIA more credit.
