GPIB allows 15 devices on one bus (and one master) with a maximum length of 20 meters. USB is 'limited' to 127 devices on one port (using hubs)
Right, but the difference is that with GPIB that works in reality, while for USB that is more or less a theoretical limit only as often even 15 USB devices won't work flawlessly together.
and can be extended to around 50 meters with simple range extenders.
Which increase latency, which again can screw up your devices
I agree LAN is better but it brings along a lot of overhead (IP stack) for the test equipment to handle
That's not really an issue these days when almost any chipset and FPGA contains at least one network interface already. IThis aside, USB isn't the most ressource-saving protocol either (esp. USB 1.x and 2.0).
and all kinds of security issues especially when equipment is running Windows.
There are even more security issues with USB, who has been the target for many attacks in the past.
LAN also isn't necessarily more insecure than GPIB, especially if the T&M network is kept separate from other networks and not connected to the open internet. This aside, Windows devices are not a higher security risk than non-Windows devices, but actually more easy to secure as it's widely known how to tighten it up, and new flaws issues get made public and (in the case of Windows versions that are still supported) also get fixed regularly. With non-Windows devices, it's pure guesswork what holes exist in a device simply because there are no regular security audits and intrusion tests, and even if a problem is found then it's pure luck if the device manufacturer can be bothered to actually fix it and provide an update for its customers. You wouldn't believe how many devices are out there that are still vulnerable to Heartbleed, and which will never get fixed (and worse, who's users will often never know that there is an issue).