The issue is that to achieve that you can't just build 90% solar plus 10% fossil fuels. You would need to build 100% solar + 100% fossil fuels for the 10% of the time solar doesn't work.
It helps that the cost of a simple cycle gas turbine power plant (before the recent data center demand spike) is around $600/kW, maybe a factor of 20 cheaper per kW than a nuclear power plant. So backing up the whole grid with such generators wouldn't be that expensive.
Good thing it's already built then! Well, of course it cost money to maintain though.
Infrastructure cost for 100% is the same as infrastructure cost for 10%? That's not true. The distribution network is the part that can't be scaled, but it can also be reused for either source, so it doesn't double in cost.
If you build batteries on the scale that the article suggests (and is probably going to happen in the real future) you can use batteries charged from fossil fuels.
It's a few percent dirtier (round trip losses) but in return you can use gas plants that are 50% more efficient to charge them rather than run peaker plants.
And of course that's ignoring wind which is nearly as cheap as solar and anti-correlated with it.