There are many reasons for the relatively slow adoption of 10GBaseT Ethernet. Here are some of them:
- Not many uses for 10G outside of the data center. For instance, 10G to the desktop is not a thing when an average 10 Megabit wireless connection is much more convenient. 10G as a campus backbone hasn’t really taken off either because stacking and chassis switches using 1G MLAGs are cheaper, even though vendor implementations are unreliable.
- You need Cat6E cable for 10BaseT, which is expensive to buy. Because of cable thickness and weight it can be expensive to install.
- 30-meter distance limit instead of 100 meters over Cat6E. 100 meters when using Cat6E shielded. Shielded cable is difficult and more costly than Cat5e in many dimensions.
- Power consumption. Coax ~ 0.2W vs 3–5W for BaseT, which matters in switches running POE.
- Coax is cheaper than BaseT because easier silicon and running coax in a rack is simple enough.
- Blade servers are still popular. I do not understand this. They need more than 10G bandwidth, so 40G is common here (again, 40G is stupidly expensive but if you are buying blade servers you don’t care about cost anyway).
- 25G is roughly the same price at 10G and volume buyers have already switched to it.
- Modern ASICs are already at 25G lane performance, which supports 10G/25G modes. No reason not to switch to 25G unless you are buying older Ethernet switches using legacy ASICs.
Don’t forget the forecast for any type of 10G is already shrinking because the majority of demand for components is determined by public cloud companies at this point in time.
The EtherealMind View
Lots of small reasons add up to a compelling view that 10G is effectively obsolete. 10GBaseT is more or less limited to use in enterprise IT where the various hidden costs are not really a concern and “What I know/what I am used to” is a critical buying trigger.