I don't read that as them "claiming 100k TPS". They identified serially having to write utxo set as the bottleneck and they estimated a high-end server could do 100k TPS on that problem.
Bandwidth is the bottle neck to scaling block size decentralized on home connections.
https://en.bitcoin.it/wiki/Scalability#Network100,000 x 0.5 KB = 400 Mbps connection at least.
http://www.nngroup.com/articles/law-of-bandwidth/a high-end user's connection speed grows by 50% per year
With internet connections in many countries still below 5 Mbps download (and consistent download rates are lower than that), that means 10 years or more to scale to that level. And to achieve micropayments a single order-of-magnitude increase over Visa scale is not likely sufficient. My guess would be 2 - 3 orders-of-magnitude within the first 10 years after launch.
Also bear in mind that perhaps 90% of the world doesn't yet use a credit card for internet purchases, so Visa scale 10 years from now is likely to be closer to 100,000 TPS.
Bottom line is crypto-currency can't scale by having every node see every transaction. Also it is an enormous
unnecessary waste of internet bandwidth and there are many other reasons such a design is not desired.