The article doesn't mention this, but they actually did make more. Or rather, they put together a new sample, VSMOW2, in 1999. It's exactly the same as VSMOW, to within measurement error. Like the original, you can buy it from the USGS. Documentation (including how it was prepared): https://nucleus.iaea.org/sites/AnalyticalReferenceMaterials/...

This actually goes both ways; VSMOW itself was carefully made (by mixing in small amounts of water from other sources) to match the composition of an earlier reference water sample, NBS-1, which came from the Potomac River (one supposes it was a convenience sample). NBS-1 is no longer used as a point of reference, since everyone's agreed to use VSMOW instead, but the USGS sells it too.