Nice in theory but the risk of crowding out remains
The model of paying upfront for the buildout of new capacity is a good idea because it allows cash-rich companies to do something with their cash other than keep it in the bank, buy back shares, pay dividends or get involved in even more risky startups. And any kind of power plant is capital intensive. The idea of dedicated power plants is also good, because it can reduce risk from the grid by not adding to the potential total peak load.
However, the effects can be the same as everyone and their dog seeks to add renewable capacity and compete for scarce resources, usually the technicians required to build and manage the facility, but could also include sites. Some of the recent deals have seen utilities sign deals to guarantee power for certain customers, potentially increasing the risk of blackouts for others. Forcing data centre operators to have their own power supply can help here but not if it means that the utilities are unable to raise the finance for their own plans, or keep the engineers to do the work. So, some kind of agreement with local regulator is essential. But when I look back at the history of US power regulators I'm not very hopeful that regulatory capture won't be one of the first victims.