I partially agree with the author. It's better if your work includes some hardware too and is not limited to software only. Just much harder for the competition to replicate the product.
But the examples he gave are all big, ambitious and universal problems to work on. How to even start on that?
I would suggest: start with some industry and learn, find what are their tools, what machines they use, what software, what are the typical problems. Try to solve some of the problems or make better tools. Find your niche. Many companies use old technology, old machines or software because they're specialized/niche and the progress of the tech is slower, less competitive in such niches. It should be easier to improve such tools, even by providing easier integration of software with the existing hardware. Much better chance of success than just trying to tackle top 10 humanity problems.
I cant really guage how serious this blog post is, either way it was an entertaining read. I thought by hardware they meant embedded systems but turns out they meant pretty much anything in the physical world.
In other words... too many people are launching software and service companies funded by retained earnings. Maybe hardware companies will require capitalization (and capitalists.) WILL NO ONE THINK OF THE VENTURE CAPITALISTS!