NETSCAPE dominated the first generation by staying ahead of the wannabe competitors by defining useful tags and getting web site developers to support them. This forced competitors to spend their time trying to duplicate the NETSCAPE flavor of HTML.
But when HTML 4.2 came out, Netscape stopped innovating and surrendered control of HTML to the standards body. And they compounded this mistake by waiting for the next standard to be finalized before starting to implement it. Microsoft didn't, they coded to the preliminary spec and thus when it was finalized they had a working product on day one.
Once Netscape lost control over the web site developers they were on the way out.
Embrace, extend, extinguish. It can be repeated. And if OpenAI isn't careful, it will be.
Contrast that with WABI, when SUN and companies tried to use standards bodies to hijack control of Windows16 API set. MS kept on evolving the Windows API set into Win32, Win64, etc unto this day. They stayed ahead of the slow moving standards committees and retained control.
The story is repeated over and over; TCP/IP vs GOSIP, WinNT vs UNIX, LINUX vs UNIX, TRUE TYPE vs TYPE1, KINDLE vs ePUB...
Tech moves fast and any player that doesn't move equally fast ends up roadkill.
Few first generation players last to the second, much less third evolution of the sector. And the lucky ones end up as niche players.
In all cases, the smart money bets on the field, not the pioneer.
And in AI, the one player cruising for a bruising is NVIDIA, for their pricing practices. They are not well-loved, even by those dependent on their wares. They need to stay far ahead of competitors because the moment somebody gets even close in raw performance they'll go the way of Type1. (A story all unto itself.)