Darwin would have liked our industry. Evolution is constant and predictable, and it's particularly so for Ethernet, from its original 10 Mbps incarnation to the current 10 Gbps version -- pricing, power consumption, and port density follow a similar trend line. But just because the evolution happens, doesn't mean that it's easy.

Aquantia's contribution is putting the entire physical layer of electronics on a single dye. Right now, the state of the art uses multiple dyes housed in a single chip, separating logic from analog line drivers, for example. The physical characteristics of these elements are very different, so putting them on separate dyes makes some sense. But it's also expensive, hard to scale, space consuming, and power consuming. As 10GBase-T moves from a few ports per switching line card to 24 and 48 ports, greater and more efficient integration is a must.

The next step is to get four or eight phy drivers on a single dye, and eventually to get entire switches onto a single chip. Once that's done, it'll be time to repeat the process for either 40 Gbps Ethernet or 100 Gbps Ethernet.

Aquantia may have a business for itself, but eventually the Broadcoms and Marvels of the world will figure out how to do this, too -- or they'll buy a startup who has. That's probably the most likely end game for Aquantia.