An 80-core processor with multi-teraflop computing power. An in-house ray tracing project. A glimpse of a future CPU powered by both general purpose IA X86 cores and a wide range of single-purpose accelerator cores. That was just one room at Intel 's recent R&D (Research and Development) showcase event over in Santa Clara, California.

Elsewhere, Intel was showing off everything from chips designed to deliver almost instant medical diagnostics, intelligent multi-radio wireless devices and the latest developments in silicon photonics. (For the uninitiated, that's mega high-bandwidth communications lasers integrated into conventional silicon chips.)

And that's not even mentioning Intel's impressive advantage in terms of silicon wafer production processes.

In short, however you slice it, Intel's technological reach is pretty terrifying. The company has research labs at scores of locations across the globe, from Europe and the Middle East to China and Russia.

Of course, you could say the R&D day is a fairly transparent effort to bamboozle journalists with sparkly new technology. Much of what Intel wheeled out is highly speculative kit which the company itself admits may not make the transition to commercial productisation.

And a fair few projects were just plain flaky - I'm not convinced that research into networking and gardening (yes, really) plays to Intel's strengths as the world's leading expert in integrated circuits, for instance.

Likewise, there's no doubting the R&D event is an effort to fluff Intel's share price. Wheeling out a huge array of impressive research projects will no doubt help to convince financial analysts that the company has the future carefully mapped out.

Oh, and keep Intel's top brass in their fancy jobs and seven digit salaries.

But even with that in mind, it's difficult not to come away impressed by the breadth and scale of the range of research projects that Intel is currently engaged in. It's not all smoke and mirrors. A quick chat with any of the engineers involved on the projects is enough to convince you of that.

They're not arrogant or pompous. Just articulate, engaged engineers who speak enthusiastically about their projects. What's more, there was virtually no beating of the Core 2 drum. No explicit reminders that Intel currently has its main opposition well and truly licked in its core PC processor market. That's confidence for you.

With all that in mind, the most obvious question is this: how the hell can AMD possibly compete? Putting aside the ATI acquisition, AMD has effectively released just one CPU design in the past five years. Meanwhile, Intel has been through the failed Pentium 4 Netburst experiment, gone back to the drawing board via the Pentium M and Core Duo and finally produced the Core 2 Duo architecture.

And now Intel looks likely to produce two parallel processor families before the end of the decade. The Core 2 road map continues with a major architectural overhaul next summer courtesy of the Nehalem core bringing goodies including a modern serial interface and integrated memory controller.

Meanwhile, the first massively multi-core Intel CPU will arrive in the shape of Larrabee, probably in 2009. Larrabee will be presented by Intel as a special purpose chip for graphics, physics processing and other relatively niche tasks.

But the unofficial scuttlebutt from Santa Clara is that the chip is actually an alternative bet on how the future of PC processing will develop. Lucky for Intel it has the resources to follow both the Larrabee and Core 2 paths.

For AMD, there is no such luxury. It simply doesn't have the resources to make multiple bets and allow time and the unfolding of events to pick a winner. It has to be bang on target with a single shot.

That's a scary thought not just for AMD, but for technology enthusiasts everywhere. It's only thanks to the competition AMD has provided in recent years that Intel has been forced to reinvent its entire processor strategy. Take away AMD and odds are we'd all be running 2.5GHz Pentium 4s and be none the wiser.