Hello!
Our first 10 Gbit/s lines dedicated only to our servers were used for the first time in Dallas, Texas, several years ago. One line is for the VPN servers and another one for the Tor nodes by Quintex. Then we had four (now six) 10 Gbit/s lines in the Netherlands. Each line was and is shared by 10 or 11 of our servers.
Then Xuange came, in Switzerland, that was the first one with an exclusive 10 Gbit/s line. Ain then followed and has been the last one at the moment.
As @OpenSourcerer says, prices in some locations (such as Tokyo) are too high for 10 Gbit/s and at least 600 TB traffic per month for a single server (2 Gbit/s 24/7 means you generate 600 TB in a month). Moreover, in order to beat the usual 1 Gbit/s full duplex, more powerful hardware is needed and a different software approach too.
Even so, on Xuange and Ain we could not manage to squeeze more than 3-4 Gbit/s (in total, up+down) when more than 150 clients are connected, and even the most powerful CPUs available on the market, running one OpenVPN instance per virtual core, suffer. The whole system get choked if we go up to 300 clients, which would be the minimum amount required to run those servers without losing money. Wireguard might help but it's uncertain and anyway many core customers of ours don't accept it for the notorious privacy problems, other customers can't use it for UDP blocks/shaping and so on, so we can't and we won't drop OpenVPN in any case. EDIT: it's not only a pure AES/CHACHA20 processing power issue, but also a conntrack and packert mangling huge queue related issue, which gets intertwined with pure encryption/decryption processing power problems. - pj
For us, the cost per user to be provided with high bandwidth is remarkably higher with dedicated 10 Gbit/s single server lines, because we experimentally see that we can not put on such a server 10 times the users a 1 Gbit/s server can handle (unless we wanted to lower the quality of service, which is not on the table). Therefore, if we want to keep the same prices and at the same time we don't want to oversell, offering an infrastructure all based on a 10 Gbit/s line per server for 2.75 EUR/month (the current price for 3 years subscriptions) is not realistic.
Remember that year after year prices of AirPVN went down or remained unchanged, and today AirVPN is probably the less expensive VPN around (ruled out the free ones, as they profile you or do worse things too).
Maybe in the future, or maybe with a different pricing, migration to all "10 Gbit/s servers" could be pursued.
We're not "over-cautious" but realistic: in the last 5-6 years, while other VPN services accumulated important debts surpassing tens and tens of USD millions (think about PIA mother company, which went down for more than 30 millions in just 3 or 4 years; and other big ones, which are forced to oversell and continuously pay for favorable bogus reviews hiding overselling in order to survive) AIrVPN never ever had debts.
Who would be interested in paying more (probably x3 or even x4) to have access to 10 Gbit/s dedicated lines (one line per server) on a wide variety of AirVPN locations with the usual AirVPN quality? We might start a survey to know.
Kind regards