Nvidia to invest up to $1B in AI startup Poolside
reuters.com
reuters.com
Amazon’s recent earnings was full of apparent round tripping.
1. Amazon “invests” in Anthropic (cost)
2. Anthropic takes that money and buys AWS (same dollars come back as revenue)
3. Amazon builds big datacenter for Anthropic (cost)
4. Amazon records a large paper “profit” because the value of the Anthropic “investment” went up after all the stuff it’s doing with the “investment” from Amazon
Meanwhile none of the above appears to actually be making any actual profit in terms of revenues > costs.
It’s bonkers. These are the same sort of shenanigans that were going on with infrastructure prior to the .com implosion. Did we learn nothing?
This will be offset against the cost of the initial investment, though?
This distorts the markets, undermines financial disclosures, and probably stifles innovation. In your example, when Amazon invests in Anthropic, do they then get voting shares and a say where the opex go?
(Rhetorical) Why are we okay with building these card houses?
1 to 1 cycle: bad 1 to 1 to 1 cycle: good?
…?
If you are a startup, and you want to buy AI infrastructure, you would typically sell equity to a VC and use that money to buy infrastructure.
If the AI infrastructure vendor would offer to buy that equity, rather than a VC, that’s preferable because you may get ongoing special treatment from the vendor once they have a stake. So it’ a great deal for the startup.
For the infra vendor- you can likely get a good deal on the investment, you get some exposure to higher upside, and it may be useful to manage your cash flow.
Seems like strategic high value moves to me. It’s no wonder the market likes it
The problem is that no 3rd party seems willing to just invest cash because they believe these companies are currently a good investment. That’s what’s raising alarm bells.
I’m not a fan of round tripping by any means. Just wondering if what you’re saying is true.
Liquidity preference.
This might be a bit too convoluted, but here is a way to get straight to the point. Assume there is a perfectly liquid asset and its nominal value is $1 million. Now imagine you have stocks that are worth $1 million, but the market price fluctuates over time.
The market price of the perfectly liquid asset is always the mean, the variance is zero. The value of the stock follows a probability distribution, e.g. a Gaussian distribution. The mean is the same, but the variance means that you can't just sell the asset at any time you want.
Even when the nominal value is the same, people prefer more liquid assets over less liquid assets. The reason is obvious. More liquid assets can buy less liquid assets at face value, meanwhile less liquid assets require you to find a willing buyer to trade, which is costly both in time and effort spent on planning the transaction. If you're impatient, you'll have to sell your asset at a discount.
The opposite is also true. If you are an investor, you prefer handing out less liquid assets first, such as credits that can only be used to buy your own products.
It seems to me that these AI companies are seeing no shortage of cash investments either.
Here the “problem” is that these AI companies don’t have enough money to buy things and the traditional pool of investors that would give cash and then sit on the sidelines has dried up. Normally in that scenario companies just go bust.
To keep throwing fuel on the fire suppliers are stepping in the “invest” so AI startups can buy the supplier’s goods. History says that’s a terrible idea.
Markets correct craziness in the end, everyone will wonder how this mess ever happened, hearings will be held, and a range of new regulations will be introduced to prevent this sort of thing from happening again. Then finance folks will come up with some new way with finance to be less boring and the cycle brings anew.
Meanwhile none of the above appears to actually be making any actual profit in terms of revenues > costs.
When Anthropic pays opex for using the data centers?
Both Amazon and Nvidia probably don't need the cash flow right now, which is why they can get away with it. It'll be interesting to see what happens if their cashflow ever runs dry.
https://techfundingnews.com/nvidia-prepares-up-to-1b-investm...
>Poolside operates across the US and Paris, focusing on coding automation tailored for government and defence clients.
>The company is also working on bold infrastructure expansion. Earlier this month, Poolside partnered with CoreWeave to build one of the largest data centres in the US under an initiative called Project Horizon. Set in West Texas, the facility is slated to reach 2 gigawatts of capacity, which is enough to power about 1.5 million homes.
Searching for Project Horizon led to this:
https://poolside.ai/blog/announcing-project-horizon
So another data center company, but focused on the Defense/Intelligence community. A hardware equivalent to Palantir?
If there's one person I don't want to lose their job to a shonky AI agent, it's Stanislav Petrov.
Nvidia has a backstop deal with Coreweave [1]. I am sure this is all above board but seeing how these giants all have incestuous relationship with each other makes me uneasy about putting money in the markets.
[1]:https://www.reuters.com/business/coreweave-nvidia-sign-63-bi...
If you invest in index funds (instead of picking single stocks), you shouldn’t be too worries IMO.
E.g. something tracking MSCI World/All-World or just the S&P500 (US-only though).
You won’t get 100x homeruns (more like 10-15% avg returns/year), but you will drastically lower your risk of losing your money.
Let me predict tomorrow’s headline: NVIDIA‘s revenue grows by yet another $1B, because Poolside just ordered $1B worth of NVIDIA AI accelerator cards.
Up until the date of the announcement, poolside was one of Amazon’s largest trainium customers, other than anthropic. This investment from Nvidia kicks one of the legs out from Amazon’s already rickety AI accelerator stool. 40 K in Blackwell GPUs is expensive even with the Nvidia investment. Poolside will likely not have the money to spend an equivalent amount now with Amazon.
Here is where I get confused:
- It does not seem very likely that all the people whose entire job it is to make these decisions are completely incompetent. Is it for some reason good to invest badly? How does this work in reality? Does it not matter if an investment goes to zero? Is every investment a good investment, because nvidia sells chips, is that how the math works here?
- Absolutely everyone and their mum talks bubble and supposedly can see it clearly (in contrast to for example the 2008 housing bubble) but nobody in power has the ability to act rationally. What the idea here?
- If it can be explained, that every investment is good now, and the answer to the previous question is something like "greed", why stop at a measly billion? Why not just, say, 10 billion? The money is there obviously. More "bad" investment = better?
Nvidia has the additional advantage that some of their invested money will make it back to them in the form of dollars spent on compute, which makes their investments cheaper for them, relatively speaking.
> If it can be explained, that every investment is good now, and the answer to the previous question is something like "greed", why stop at a measly billion? Why not just, say, 10 billion?
If the startup isn't looking to raise 10 billion, why give it 10 billion? Instead, give 10 startups one billion each, to spread out the risk.
I am not much of a trader, but I assume there are always insane opportunities to invest a few billion dollars in high risk/high gain stuff, no? Why wait for AI? You can try to time any market. What explains the enormous amount of money into this one?
That's fine... I guess. Somewhat boring. I have no clue why we would feel the need to dress it up and make it sound like insight into the market.
It is the same psychology behind meme coins. Everyone knows they are worthless, they gain value because everyone is trying to make everyone else the greater fool holding the bag.
2008 was less a bubble of irrational exhuberance and more a bubble of overvaluation enabled by fraud of bundling bad mortgages into A rated investment products
I think the properties of a true bubble mean that it is not irrational to invest in the bubble and that is a part of what causes the bubble. If your neighbor makes $100k flipping a house and has no special skills above yourself then it is not irrational to also try to flip a house.
This bubble especially has that property. Like Zuckerberg has said there is a big risk of not taking enough risk on something like AGI or super intelligence. That is a perfectly rational position but that is exactly the self reinforcing nature of a speculative bubble.
I also don't think in general a bubble is fundamentally hard to spot. It is hard to profit though from the bubble bursting because of the self reinforcing process at play that you have to bet against.
As long as it could still be far worse to not take a risk, then it's not actually a bubble, right? It's just an early investment. If I have to do something now, to not fall off the track, then the space might not be super clear right now but clearly still on track (because otherwise the correct strategy is waiting for the bubble to burst and maybe then invest).
Or maybe we are just really unclear about what "bubble" actually implies. To me it's the point at which a lot of people have invested a lot of money into something, that will be worth less, than that total amount of money. Is that wrong?
The Data Center Bubble: https://www.brethorsting.com/blog/2025/10/the-data-center-bu...
Global Crossing Is Reborn: https://pracap.com/global-crossing-reborn/
Nvidia's larger investments are even more conditional. With OpenAI, the investment is contingent on building huge data centers in the future. If the bubble ends, Nvidia pays nothing; if it continues, Nvidia has locked in a customer.
Nvidia plays smart with little risk.
A company involved in round tripping passes fictional money in a circle and every company that touches it claims both revenue and expenses simply for passing it along.
What does this mean? What's the trick?
Well we can argue on the meaning of trick I guess.
Share buybacks are essentially a way to achieve the same effect as dividends, but in a non-obvious way, which has the benefit of avoiding taxation. That's a "trick" in my book, but I guess terminology doesn't matter that much.
> They'll still pay tax on it when they sell it.
Not but it's not _equivalent_. The tax paid on capital gains is not the same as the withholding tax. And paying tax _after_ compounding is not the same as paying it _before_.
Share buybacks are _effectively_ a trick to circumvent withholding tax for investors not willing to divest.
But why? Minimising tax legally is...legal, and not a trick. That's all tax avoidance is.
> The tax paid on capital gains is not the same as the withholding tax.
That seems totally fine - if the rules are different then that's up to the people who write the rules. It's not a trick to choose to be paid via one method or another.
I think we are lost in translation here. I am not a native English speaker, so there may be a subtle implication in "trick" that you see and I don't.
I meant "trick" as in "trick of the trade", a clever/crafty way of achieving something that may not be obvious for less experienced individuals.
Re-phrasing my original comment for clarity: "Share buybacks are just a technique to lower WH tax, why do you see this as anything related to round tripping as related in this article?".
The trick is essentially to buyback your own shares and destroy them. That effectively redistributes the value you bought to other shareholders, much like a dividend would.
How is that better you may ask? two reasons:
- Most investors prefer to accumulate rather than receiving cash. If you post dividends, they are immediately subject to withholding tax, so you get taxed before reinvesting.
- In a lot of cases, capital gains tax and withholding tax are different, the former being much lower than the latter. This is especially the case for funds with foreign UBOs, which incur 2x15% WH tax at the source.
- Buybacks are just more flexible, those that want cash can sell, those who prefer to accumulate are happy to stay, there's no real downside.
Some of this is paying for barely useful assets using inflated stock, or with cash borrowed with inflated stock as collateral for the cash.
Slightly offtopic: If a company does a stock buyback because they think its undervalued, what happens next? Does the stock go up and they're satisfied? Does the stock go up and then they sell it?
If they're selling it to realize profits, I say that it tantamount to pump-and-dump. If they sell it just to hike the price, why not distribute dividends with their excess cash reserves?
The reason this is often done instead of a special dividend is that dividends create an immediate taxable event for all investors, which is considered less flexible than the capital gains tax associated with a stock buyback.
Besides the tax treatment difference, it's mostly a signalling/communication choice: share buybacks increase EPS which is a nice story, whereas dividends signal reliable profits.
> "It is the French equivalent of what the U.S. announced with Stargate. It is the same proportion," Macron said.
As much as the whole AI thing feels like a bubble, this feels less round-trip-like to me than some of the other recent deals which have been in the press.