A sunshade is a feasible direction.
And it has two distinct variables - what to make and how to make.
Those variables define the cost and feasibility, and possibly the location and potential complexity of implementation so as the usefulness of the end solution(as an example use that energy in space for something useful)
In 2016 China announced some plan worth 50 trillion to "displacing fossil fuels" to the 2050 - so 5 trillion is 10% of that, which is relevant to any country on the planet, this or another way.
Simple solutions like just foil like shreds may affect the situation not necessarily in a good way, or effects may be different for different countries and that is one of the problems of simple solutions, the lack of flexibility of real-time adjusting of the system. So it more theoretical solutions which show some direction, but not necessarily is an implementation worth pursuing.
But 2% of energy which we need to block and potentially can use in the L1 location. It is a lot of energy(it's huge - with something like 40% efficiency 2% it is about 1380TW). To build an equivalent source of energy here on earth, with price 1B per GW it is 1380 trillion. So if there is some way to build something more useful there in L1 for a fraction of that cost, it means such a solution not only can negate certain aspects of warming problem but also deliver some services to earth economies as a 1380 trillion worth installation.
In regard of combining solving one issue with some useful work, ServerSky looks like an interesting proposal in the context. Considering Generative design such a computing power could be used to address designs so as processes of production and potencially distribution(more variables more complex is the problem and it is essential enough problem to consider such exotic ways to address it(to understand the ways of solving it) like this one Slime design mimics Tokyo's rail system: Efficient methods of a slime mold could inform human engineers)
So as more wide and profound use of deep learning algorithms for different tasks, which are on the rise today, and when teaching them is a somewhat expensive problem and takes week's and more to accomplish on cloud systems, and costs significant sum.
So connecting solving one problem with what may be a solution for some of our demands, it may allow to look differently on the question what is the reasonable price, and scale of efforts worth taking, and complexity of the task in terms of R&D.
So changing a focus from some issue which is a spot on the picture and to think about the situation more broadly, thinking not about patch solution but about creating a useful system, may provide us with an understanding of opportunities in the case.
All that to say, that in some cases it worth to think about the moon as a palace which may help to implement such change(which btw produces other options to solve/fight some consequences of GW - acidity, sea fertilizing etc)
But in the case assessing the price of a solution is a quite complex topic, and obviously classical ways of establishing some infrastructure on the moon on the scale required, may be too expensive. Classical means bringing all the production equipment form the earth. But there are options in a direction like that Affordable, Rapid Bootstrapping of the Space Industry and Solar System Civilization - but one of the problems defending feasibility here also is a problem, but it rather feasible.
And if we may consider investing 5trillions in something(meaning it is a practical sum), it can be more than enough for developing and implementing of quite complex solutions in space. Our today's achievements in automation, in electronics and software they allow us to implement different ways to achieve the sunshade solution - with an ability to provide service and indirectly and directly affect how we do production and designs on earth - that indirect(and direct) positive effects on economy from space programs NASA talks so often.
Using moon may reshape the problem, in different ways, but in terms of delivery, the most obvious changing the problem from Earth to L1, to Moon to L1 which allows thinking about different massdriver solutions without a stretch and the problems earth atmosphere creates for those solutions.
The topic definitely needs more public discourse and more consideration than just theoretical direction, and it can be that game-changing space program we all would like to see. There are problems to solve, but they also bring different solutions which are worth to pursue.
Unfortunately, estimations at this point are quite useless, I could probably try to defend 1trillion investment, but also 20 billion do not look for me like a joke, it just much harder to defend as it needs more details as a plan/road to be able to do so. 100 billion may be a realistic and defendable position, which may give much more than just a sun shade.
I think it is an important problem or a change in a perception of the problem, so you may ping me on discord, if you have some questions or wish to discuss it.