On Tue, 22 Jul 2003, Allen Akin wrote: > (Sorry for the delayed reply -- I've been travelling.) > > On Sun, Jul 20, 2003 at 04:15:51PM -0700, Mark Vojkovich wrote: > | On Sun, 20 Jul 2003, Allen Akin wrote: > | > ... It may be that there *aren't* any million-dollar > | > per-employee revenue opportunities available in the market. If you > | > want to expand, you have to take the best of the opportunities that are > | > available, even if that means revenue-per-employee goes down. > | > | Correct, however, graphics hardware companies have alot of tangential > | markets they can go into that hold large revenue potential. For example, > | when NVIDIA needed to grow, it didn't try to collect crumbs with software > | opportunities for existing hardware. It got into the core logic business > | with nForce, entering itself in a whole new market. ... > > I don't disagree with the fundamental point you're making, but is nForce > really a good example? There are lots of highly-integrated chipsets out > there. It may be that nForce succeeds primarily because it offers > excellent graphics (NVIDIA's main market strength) at a lower price > point, rather than something wholly new.
You'll have a hard time finding reviews of the internal graphics. The interest, at least as far as the press is concerned, seems to be almost totally in the chipset performance. nForce2 is available without internal graphics. If I recall correctly, nForce3 (I believe there were some reviews out today) has no internal graphics. > > | the emerging opportunities for graphics devices: new markets in settop, > | palmtop, cell phones, consoles and other appliances, I get the feeling > | that there are so many markets left untapped, yet not enough resources > | to devote to them all. ... > > Perhaps that's a good reason for NVIDIA to be active (in some way) in > open source efforts. Shared development cost means lower barriers to > entry and higher return per employee. The trick is to get the > functionality you need from the shared development effort, while > avoiding the "free rider" problem. How to do that is worth further > discussion. > > | ... I don't expect the current graphics hardware > | companies to ever be in a position where they have to scrape the bottom > | of the barrel for revenue. > > Accepting less than $1M/employee/year is hardly "scraping the bottom of > the barrel." :-) > > But seriously, we're in a good period for graphics market development > right now. It probably won't always be this way. Unless a monopoly > develops, eventually the market will mature, the hardware will become > more commoditized, and margins will go down. The vast majority of > design/manufacturing companies in the world live with lower return than > the top-tier graphics hardware vendors. I do not expect graphics hardware to become a commodity. It is something that graphics hardware vendors fight hard against, and commoditization has not been the trend. > > | Another thing to keep in mind is that users shifting from > | Windows to Linux is not creation of a new market per se. The > | total hardware revenue does not increase. It merely creates more > | work for the vendors as users shift from one platform to another, > | and prevents vendors from being able to focus on one platform. > | Hardware vendors have little incentive to encourage this > | change. ... > > There is at least one very significant incentive -- driving down system > cost and enlarging the market by commoditizing the software This is something graphics vendors will fight hard against. They don't want to commoditize the software. This business is about the software value add as much as it is about the hardware. The goal is to have your graphics kill your competitors, and that is done by preventing commoditization of software and hardware. It involves outfeaturing and outperforming your competitor and protecting your software and hardware IP. > (http://www.joelonsoftware.com/articles/StrategyLetterV.html). This is > much harder to do in the general-purpose desktop space than it is in > specialized and embedded systems, so I agree that it's less likely to > happen (or will happen much later) for desktop systems. But it's an > important factor in other markets. > (http://www.linuxdevices.com/news/NS4980264574.html for one example.) > Mark. _______________________________________________ Devel mailing list [EMAIL PROTECTED] http://XFree86.Org/mailman/listinfo/devel
