The limiting factor is the bandwidth of the people making the investment decisions. There's no way to manage a portfolio of that size and dedicate the amount of time you'd want to any one of them.

Also, why give $1 million to 4000 startups when you can put more wood behind the arrow of your best ideas? The point of active management is to generate alpha, doing so requires superior analysis or some other "edge".

How do you know what your "best ideas" are? And, are they better than the average of those 4000 startups? Are they better than the top 1% of those 4000 startups?

It's not clear to me that, starting from here, NVidia is going to be a fast-growth company - especially not compared to a group of startups.

its possible they are using NVidia as a proxy for a bet on Machine Learning.

This is a bet on AI and machine learning. Not video games. NVidia is currently the publicly traded company that best fits that narrative. If you think it will get big, this is the company you buy.

> the bandwidth of the people making the investment decisions

why do they need to make any decision at all - if it's like lottery, why not invest in all 4000, let it run wild? The survivors after their money runs out are the good ones.

Presumably bad ROI

That's a big presumption isn't it?

It was accepted wisdom that you had to pick individual stocks until Vanguard popularized the index fund and now it's accepted wisdom to do the opposite.

It's one thing to pick up stocks of market leaders (which may or may not be having a good time this year around) and another thing is to pick stock of noname startups from the bottom of the pile. Far more risk.

There's a reason early stage VCs do dd, it's to weed out complete garbage which wouldn't ever possibly do any return. Probably it is not worth scaling that and easier and more measurable to just stick such big amounts of money into later stages.

When I was at Goldman (I left years back), they were pretty proud of the 10k small businesses scheme. They ran it in both the US and UK.

Not sure how much capital each got, but I guess it's similar to what you were thinking.

http://www.goldmansachs.com/citizenship/10000-small-business...

I'd say most Stanford, Berkeley, or MIT grads can easily land a few hundred thousand dollars (which is honestly more than enough to produce something or test an idea) with any half-decent idea or half-decent ability to bull-shit. Most who fail do so because they didn't pass the passion bar to bull shit or push something out.

I assume you mean $1 billion ($1,000,000,000 / 4000 = $250,000) ? $1 million to 4,000 people is only $250 each.

I would also assume the risk would be lower to give 2000 people $500,000 so they could hire people ($250,000 is only good for two-four person teams for maybe 2 years? Excludes hardware/travel if they are building something other than software products.)

I believe he meant $1,000,000 x 4,000 = $4 billion

I think he means take the $4bn invested by Softbank and split that into 4,000 separate $1m investments!

If they do that on Monday, I am opening 100 startups on Tuesday... even if I think most of them may go down...

Dave McClure at 500 startups is about the closest model to that