Yes, coax can be manufactured at 100 ohms. However, it wouldn't be suitable for building an HDMI cable because HDMI signals are run balanced, which requires a symmetrical cable architecture (which, in effect, means twisted pair). It would also be possible to run HDMI by using two 50-ohm coaxes for each balanced line, grounding the shields in common to that pair's shield ground. To do that, though, you'd have to really control the lengths well because skew is a major factor in keeping HDMI signals together.
I have spent a lot of time with engineering staff at Belden on this problem, and I would point out that Belden is, among other things, a world leader in super-high-frequency data cabling, and has a whole range of Gigabit products out. I've seen sales samples of Belden 10-gig cable, though I think it hasn't hit market yet. And you know what? Every last one of the engineers I have ever heard comment on this issue, without any prompting from me, has observed, "it's a shame they didn't just run this in coax." When the guys who know as much as anyone in the world about running high-speed data in twisted-pair cable tell me that, I suspect that my making this point to the world in this article is not exactly "misleading."
Other high-speed data technologies aren't as fragile as HDMI. They are generally not run parallel; they are run with error correction. The fact is, and anyone who's had trouble running this stuff over distance will understand: HDMI is an unreasonably fragile interface, and the problem could easily, and cheaply, have been avoided.
SDI, as is used in the broadcast industry, would not be the only option for consumer use. HDMI could have been done more or less as-is, but instead of four parallel balanced pairs, four 75 ohm coaxes could have been run. Very low attenuation, excellent timing to minimize skew, excellent impedance control to limit return loss--it would simply work better. And if one needed to run it over extreme distance, a cheap breakout box with four BNC jacks and a run of CAT5 to carry the low-speed data and miscellaneous what-not would be all you'd need to extend it considerably, with no active circuitry. Using RG-11 cable you could run it 400 feet with no booster and remain within spec for attenuation. And that likely means you could go quite a bit further, failing spec, but still functional, in most applications.
Now, hardly anybody wants to go 400 feet. But with twisted pair, 50 feet has been a challenge for a lot of manufacturers. And if your cable works today, at 50 feet and 1080p, there's no guarantee that it'll work tomorrow at 1080p/12 bit color, 1080p/16 bit color, or whatever else the next generation of the HDMI standard might throw at it.
Kurt
Blue Jeans Cable