solyentcola is actually correct. Google Wallet 1.0 used tokenization, but it was issuer side, not network side. When the networks weren't interested in tokenization, Google Wallet relaunched with HCE.
As far as I understand it, they were issuing an actual credit card applet to the secure element for the first version, and later switched to the "proxy card" model and HCE.
Issuer-side tokenization wasn't even launched until 2014-09 by both Mastercard and Visa ("coinciding" with the launch of Apple Pay), so they couldn't have used it for the first version of Google Wallet.
Mastercard and Visa launched network-side tokenization, not issuer-side tokenization. And yes, it used a Mastercard PayPass applet, as does network-side tokenization.
When it comes down to it, tokens are just credit cards that have a mapping back to an account. The network-side tokenization spec includes additional measures for provisioning and security, but they're still just credit cards. (If that weren't true, then all of the merchants would require terminal upgrades to work with them).
In issuer-side tokenization, it's the issuer (the user's bank/cc company) that creates the token, and they map it to accounts internally. In network-side tokenization, it's the network (MC, Visa, etc.) that handles the token creation, and they map it back to the previously issued credit card before the charge goes back to the issuer.
