Given xAI built its 100k gpu datacenter in a very short time, is the infra really a moat?

I'd say it is because the $ it takes to build out even a small gpu data center is still way, way more than most small cos can do. It's not an impenetrable moat, but it is pretty insulating against startups. Still have a threat from big tech, though I think that will always be true for almost everything

I don't think the hardware is that easy to source just yet. Musk pulled some strings and redirected existing inventory and orders from his other companies, namely Tesla, to accelerate delivery.

xAI does not have infra to sell the service and integrations of it to enterprises and such. It's an open question if "models" alone and simple consumer products that use them are profitable. So, probably hyperscale cloud platform infra is a moat yes. Microsoft has Semantic Kernel, Microsoft.Extensions.AI, various RAG and search services, and an entire ecosystem and platform around using LLM's to build with that xAI does not have. Just having a chat app as interface to one's model is part of the discussion here about models as commodities. xAI does have X/Twitter data which is a constantly updating source of information so in that aspect they themselves do have something unique.

they likely utilized expertise/supply chain from Tesla

Shouldn't this problem self-regulate, though? Ultimately, investors mainly care about the returns and if you can get better returns elsewhere due to these fees, they will switch.

If they can charge large amount of fees and still stay competitive, then good on them, right?

There's a lot of friction. You won't switch based on one year, which would just leave you chasing last year's lucky winner (who will likely revert to the median next year). It takes a long time to realize that your hedge fund is a loser.

The whole point of a hedge fund is for you to let someone else do the worrying. So the market is decidedly inefficient.

> It takes a long time to realize that your hedge fund is a loser.

it's been many decades since the existence of statistical analysis of hedge funds (as an aggregate) that demonstrates their lack of edge over benchmark passive index funds.

Some hedge funds would still out-perform. Most don't, and those who do tend to charge fees up to their level of edge, and leave only index-benchmark returns for their investors.

If you don't heed this evidence now, you deserve to lose money to these funds.

The purpose of a hedge fund is to ‘hedge’ ie deliver alpha and not index beating returns. Obviously the fee issue is another layer.

That was the original idea. They realized that they could also run the risk equations in reverse and make index beating returns, at least in the short run, and pretend that they could also do it in the long run. Hedge funds got a reputation as overpowered mutual funds for the wealthy.

It's kind of like buying lottery tickets from a vendor and when you win, but he steals the winnings, you switch to another one.

Except, also, you don't actually ever see the tickets.

> Ultimately, investors mainly care about the returns

Not quite. It also matters how and when returns are generated. Some vol funds make 1–3% a year on average, but they still manage billions because when markets crash, they (presumably) crush it — and that’s when investors need them the most

Depends on if there’s capital gains tax when leaving the firm.

Perhaps. But HF capital is usually locked down for a few years. So there is some friction to switching

True, but I believe (although last explored this question a long time ago) that the majority of the capital is not locked.

Quick search shows current lock ups range from 6mo to 2yr. There are also some funds that have withdraw limits and short windows for doing so (I.e. once a quarter perhaps), but of course these tighter restrictions are only relevant for the funds that have the clout to implement them

You are right about Social Security. It only holds Treasuries. But, there are many other pension funds like teachers pension, government workers pensions, etc that manage working class money.

Am I getting this right?

  - LSD SQL is a way to query a webpage to extract data...sort of like BeautifulSoup
  - You can "give LSD to Claude" using Claude's model context protocol
  - Now Claude can use LSD to extract more relevant data from webpages and use it to answer questions

Yeah, looks like they're trying to gamify AI... nah.

YouTube ads generated almost $9B last quarter. I think that should be a viable business on its own?

Do you mean bitcoin/crypto or did you have something else in mind?

I think most people don't really know what those things are, or what the fundamental concepts are, so I recommend people research decentralized technologies that are related to money from scratch.

I've decided that browser extensions are too much of a security/privacy risk. I just stick with 1password extension and an ad blocker extension that uses Safari's Content Blocker API only.

And then from time to time I have a dedicated profile on Chrome to use other extensions that might be useful, but I don't do day-to-day browsing there.

I'm similar

I have all relatives set up with a separate browser for e.g. banking, and it has no extensions at all

then the usual internet browsing one with the security nightmare than is the chrome/firefox app store

What can go up so quickly can also come down quickly.

It's impressive that Nvidia has been delivering explosive growth in their earnings to match their valuation, but it is valid to be concerned whether the world can keep investing in GPUs like this sustainably.

How do you do this with other DBs?

The new Variant type exists independently of JSON support, so it seems good that they handle it properly.