In the late 1990s, when I was working on physics engines for animation, I was talking to a major Hollywood director. He'd done some of the first films that had both live and photorealistic CGI characters. He wanted to get the cost down, so he could make $20 million movies. At $20 million, he could direct; at $100 million, he was running a huge operation that had to have everything pre-planned in great detail.
His model was the early CGI cartoon, "Reboot". Reboot was a weekly half hour cartoon made by a staff of about 30. He wanted to get to that level of productivity at theater quality - make a 2 hour film in a month with 30 people.
That didn't happen. Not even close.
It's been tried. "Sky Captain and the World of Tomorrow" started as a low-budget picture rendered on Macs. In that film, if nobody touches it, it's CG. Ended up costing $70 million. Worldwide box office $50 million. Fail. "Iron Sky" was made for €7.5 million. Worldwide box office $11.5 million. Fail.
There were successful low budget directors in the past, Roger Corman being the prime example. (His autobiography is titled "How I made a hundred movies in Hollywood and never lost a dime".) It's harder to do that today. Viewers today expect incredible production value. Most TV shows have more production value today than 70s movies did. Hollywood has so many people on set because they bring a team together on short notice to do a job, then disband the team. They need many competent people with different skills to make that work. If you cut corners, it looks like Youtube crap.
On big movies, we've mostly replaced set painters and carpenters with people who sit at workstations and do the same job with CG models. "Big" is now cheap, but "detailed" remains expensive. Procedural visual content generation can generate good landscapes and vegetation now (check out SpeedTree), but as yet, nobody has been able to procedurally generate one convincing block of a city street seen at ground level. Making GTA V cost $265 million. Not seeing an incoming reduction in production cost.
Netflix has no huge advantage. HBO is in the same position - they make content to sell to their own customers, and know exactly what sold. Data collection is retrospective. Trying to figure out what movies will be box office failures in advance remains hard. That's why Hollywood generates so many sequels - predictability.