In the coming years, I expect that companies that can develop fast, power efficient graphics processors will have a leg up as the world of computing -- both client-side and server-side -- evolves.
Chip-maker Intel (INTC 2.75%) has been building graphics processors for a long time, and has even been selling graphics processors integrated into just about every consumer-grade processor that sells for more than half a decade.
Image source: Intel.
Moreover, Intel has even shown a real willingness to experiment with offering higher-performance integrated graphics configurations in a bid to drive up its processor average selling prices (though its most recent effort to "go big" didn't work out so well).
At the end of the day, though, I don't think Intel's current graphics technology is still "leadership" in any sense of the word -- performance, power, or area.
I think that a big part of the problem here is that while Intel is giving its graphics technology more attention, it's still a second-class citizen compared to the CPU cores inside of Intel's products -- at least, that's what Intel's actions have signaled to me.
To that end, I'd like to propose a potential way that Intel can dramatically improve its graphics technology.
Enter the stand-alone GPU market
These days, high-performance stand-alone graphics processors are exciting. Just look at the substantial rerating that NVIDIA (NVDA 1.29%), which is the world leader in high-performance graphics processors for gaming, professional visualization, and data center, has gotten over the last couple of years:
NVDA PE Ratio (TTM) data by YCharts
Not only does NVIDIA stock command a substantially larger multiple than it did before, but the company's raw profits have never been higher:
NVDA Net Income (TTM) data by YCharts
Given that Intel already spends significantly to develop the graphics technology to integrate into its PC, Internet of Things, and even data center chips, the company may as well make graphics a true first-class citizen and try to build power/performance competitive stand-alone graphics processors for the professional visualization and gaming markets.
At best, developing and marketing such products will allow Intel to cultivate additional, relatively high-margin revenue streams by building products that can fundamentally take advantage of Intel's proclaimed manufacturing technology leadership.
Even if Intel doesn't rake in billions in additional annual revenue from such products, the work that goes into trying to build competitive graphics technology won't go to waste; it could have the effect of wringing out excellence from Intel's graphics technology development teams, which would in turn make the graphics processors integrated into the company's processors (these chips literally bring in tens of billions of dollars for Intel each year) more competitive and valuable.
I don't just mean that Intel would develop better graphics hardware if it were to jump headfirst into the stand-alone graphics processor market (though I believe it would have to; performance and power efficiency are key in these markets), but it'd be forced to make dramatic improvements in the driver software that's responsible for getting the most out of that hardware.
To attack the gaming and professional visualization markets with stand-alone solutions, Intel would need to significantly bulk up its driver development teams to help wring the best performance out of the latest professional software as well as the latest games.
Since the graphics technology integrated into Intel's processors would share the same DNA as the hypothetical stand-alone graphics processors, software/ecosystem improvements would benefit Intel's entire graphics-enabled portfolio (this includes PC and Internet of Things chips).
Thinking long-term
If Intel were to go down this road, it'd be a long one -- one that requires significant tenacity, even in the face of elevated near-term operating expenses with little to no revenue to show for them.
However, in the context of Intel's $12-billion-plus in annual research and development spending (and remember, that spending already includes spending toward the current integrated graphics technology), I don't think the expenses would be that elevated.
Note that NVIDIA -- again, the unequivocal market leader -- spends under $2 billion annually in research and development (though the figure is growing). I can't imagine the incremental spend for Intel to really take graphics seriously and try to go after some of the markets where stand-alone graphics processors are hot would be too much for Intel to handle.
It's a question of whether Intel thinks such an investment would ultimately be worthwhile. I think it would be, if only for the side-effect of substantially improved integrated graphics technology. Actual financial success in the gaming or professional visualization stand-alone graphics card market would be gravy.
