Oracle does not have any proprietary IP that OpenAI depends on. That means Oracle is providing commodity services to OpenAI.
AI hardware is advancing rapidly. If NVIDIA's next generations deliver radically more tokens per watt and per dollar, an Oracle facility designed, financed, and contracted around an older GPU generation can be overtaken before it is fully online or fully monetized.
That's a big risk.
And to finance this they are making deep cuts to the ERP business where Oracle does have proprietary IP.
> If NVIDIA's next generations deliver radically more tokens per watt and per dollar,
> an Oracle facility designed, financed, and contracted around an older GPU
> generation can be overtaken before it is fully online or fully monetized.
Every user of Nvidia's gear has thought of this, and Nvidia provides a solution: Newer components are (usually) compatible with older racks, and Nvidia will buy back parts you replace with newer hardware.
For example, let's say you have an NV72 with Blackwell-ultra. Rubin comes out, and you decide you want to upgrade. There is a SKU of Rubin designed specifically to work in the racks shipped with Blackwell. Use it, and you only need to replace the compute board (not the power system, cooling, networking, etc.)
Nvidia will buy back the Blackwell boards you remove (and sell them to someone else, with a warrantee).
If you want to do an upgrade that increases power use, you may have to get a new rack. All of the parts are modular, and all parts can be sold back to Nvidia.
Nvidia does this because they have to. Without it, many customers would consider other vendors when upgrading, or delay purchases in case something better comes along later.
I'm not really sure they have thought this through. Or at least their solution has been to essentially insulate the company and even the physical building from the debt used to buy the GPUs. These SPVs typically lease the building and secure the debt to the GPUs themselves so if it all goes pear-shaped, it'll be the debtholders who are left holding the bag, priamrily.
But I have questions:
1. How often do GPUs fail? I've seen figures that this is as high as 9% annual failure rate for DC GPUs;
2. How easy is it to get replacement stock? At what cost?
3. If you trade in, what value do you get for your GPUs?
4. What is the availability like? How does NVidia decide who gets the next wave of GPUs? That's NVidia's big power here. It's a bit like how Boeing and Airbus decide on customer order and if you upset them, you may lose other orders. This happened to Qatar Airways over the A380 [1].
5. How long are the current form factors and interfaces going to last? How long before the power requirements get larger than what DCs are currently built for? Blackwell GPUs use a lot more power than, say, 2 or even 1 generation ago did.
GPUs are a huge cost factor, allegedly more than 50% of the cost of a data center [2]. If you take a bath on the trade in and are forced to do it because you can't compete with the current generation's performance-per-watt, that's not a good situation to be in.
Sometimes I wonder if the people committing and spending this much money is really thinking about this or not?
My feeling is there is some kind of upgrade packaging done here. Like if the chips advance by X factor it can be offered at a massive discount or something? Otherwise it seems so silly to buy so much hardware which is evolving so fast.
I'm no fan of Oracle, but you could make the same claim about land developers, or datacenter operators, or REITs. Or a random conglomerate like Unileaver or GE or Siemens.
Ultimately, someone has to provide "commodities", and if you think your ship is the right vessel to do it betting hard on this is some kind of strategic direction.
Whether they are right, and to what extent debtors and the market will put up with cyclic and performance issues, etc.. I certainly won't hold the stock, but I wouldn't have earlier either.
ERPs are done. Data center permits, space, contracts to buy scarse GPUs and energy are the moats today. The chips can provide more tokers per watt but the world needs exponentially more tokens each day
Curious to know why you think ERPs are done. I actually do somewhat agree with you in that their best days are behind them.
It is going to be incredibly difficult to bill outrageous amounts for armies of consultant to implement or customise the product when agents can do the same for much less, or at-least a 5-10x reduction in headcount for similar complexity of changeset.
If anyone is in the industry, please feel free to chime in as to what you are seeing on the ground.
>Oracle does not have any proprietary IP that OpenAI depends on. That means Oracle is providing commodity services to OpenAI.
Huh? Is Oracle's plan with this funding to provide compute services for the big AI specialist companies? I assumed it was a buildout for their own AI services, as you're right, the former wouldn't make sense. (At risk of making a trollish, low-effort comment, OpenAI/Anthropic are savvy compute users and would have to be insane to risk getting entangled with Oracle.)
I read the article and I couldn't tell from it which of these is the case.
On strictly business terms, you're absolutely correct. Considering that OpenAI and Oracle both eat from Trump's hand though, I think that Larry Ellison has a few tricks up his sleeve to remain a "necessary" contractor. This is probably a shorter-term investment to pad out the coffers 5-10 years from now.
If rampant despotism is the end-goal, Oracle has already aligned themselves with the despots.
A company that is purely in Denial, even doubling down at the risk of toppling the whole financial system should be placed under the financial equivalent of arrest and its leadership actually arrested. They're obviously playing with fire and it's alarming that no one's in the administration is batting an eye.
Oracle not able to provide its usual "services" would disrupt many institutions which unfortunately rely on their technology and assistance operating the often expensive and complex Oracle integrations they've set up, for better or worse. I really despise Oracle and the Ellisons and they don't need a bailout but I'd hate to see what could come from many significant collapses of technology for social institutions.
If Oracle suffered a massive financial collapse wouldn't all those aspects of its business that are needed and paid for "just" get sold off and be managed as a new entity? I quote just because it would be an enormous amount of paperwork and organizational restructuring, but people with enterprise scale money tend to get to keep the services they pay for.
This came up in the last discussion: The TL;DR is that shareholders go to zero, bond holders take a haircut and then own "new Oracle" (Opacle?) that can generally sustain itself with revenues from agressive JDK audits, RDBMS licenses, and ERP/PeopleSoft.
It's not dilutive -- current shareholders get $10B (or 97% of $10B, whatever). Market reaction is some combination of profitability concern / bankruptcy risk because Oracle is super leveraged and taking on even more debt. Maybe the AI hyperscalars will slow down spending in 2028.
Yes, even if the company raises $10b, the ownership stake and voting power of a share is reduced by 2.5%.
And if that $10b figure is a fixed target, then if the share price continues to fall, the more shares they will need to issue, resulting in even more dilution.
Previously, I thought investing in AI companies would be a good hedge against AI taking my job, but I hadn't considered the possibility that they fund the build out by issuing shares. Is there no reprieve?
The shares you own (usually) give you voting rights. Whether that makes any difference is another story.
But I'm not sure what people think the stock market is for? It exists primarily so that companies can access capital markets namely by issuing their own shares.
Historically for companies that were growing rapidly or had large capital needs, private markets weren't large enough to handle that. Times have changed, but the whole point of going public is to be able to raise capital by issuing shares.
maybe not. There could be mass unemployment and all the big dogs could also lose from the buildout while smaller players (non-public) take over. It could be a lycos / yahoo / google situation where there was no "winner" to invest in during the initial dotcom crash, and there was large-scale disruption all the same.
I'd like to note that "the sector that employs you" should be considered in diversification planning.
A stretch of unemployment can be weathered with good investments, or a crashing market can be waited-out by staying firmly employed, but both simultaneously is what forces people to "sell low".
I think they are actually underpriced relative to "taking away" jobs. I personally believe ant at 1-2T is pretty reasonable if you consider claude to effectively be the "best devtool of all time," aka it literally writes all software for you.
I think (somewhat arbitrarily), to get to higher valuations, you are going to have to disrupt more than just the software business. Which is what most labs are trying with finance, biology, chemistry, law, etc. That might be enough for a few hundred billion in TAM more.
I think to get to a ~5T+ valuation, then we have to start thinking about AI meaningfully displaces classes of jobs, maybe in the single digit %'s. As in, _global_ SWE headcount starts going down, but revenues for software companies as a whole go up.
I think for ~10T+, you definitely need meaningful job displacement across most white collar sectors.
Things I hope for:
- Oracle defaulting on its debt commitments.
- Anti trust action into Skydance forcing a breakup.
- Larry forced to cover the difference by selling family silver.
- Lanai going back into the hands of the people of Hawaii.
Just for the laughs and for forcing me to bear with some of the worst applications of all time.
> - Lanai going back into the hands of the people of Hawaii.
Lol, it's 2026: it'll just go to some other billionaire. Probably Marc Benioff. I heard he likes owning Hawaii.
Oracle does not have any proprietary IP that OpenAI depends on. That means Oracle is providing commodity services to OpenAI.
AI hardware is advancing rapidly. If NVIDIA's next generations deliver radically more tokens per watt and per dollar, an Oracle facility designed, financed, and contracted around an older GPU generation can be overtaken before it is fully online or fully monetized.
That's a big risk.
And to finance this they are making deep cuts to the ERP business where Oracle does have proprietary IP.
> If NVIDIA's next generations deliver radically more tokens per watt and per dollar,
> an Oracle facility designed, financed, and contracted around an older GPU
> generation can be overtaken before it is fully online or fully monetized.
Every user of Nvidia's gear has thought of this, and Nvidia provides a solution: Newer components are (usually) compatible with older racks, and Nvidia will buy back parts you replace with newer hardware.
For example, let's say you have an NV72 with Blackwell-ultra. Rubin comes out, and you decide you want to upgrade. There is a SKU of Rubin designed specifically to work in the racks shipped with Blackwell. Use it, and you only need to replace the compute board (not the power system, cooling, networking, etc.)
Nvidia will buy back the Blackwell boards you remove (and sell them to someone else, with a warrantee).
If you want to do an upgrade that increases power use, you may have to get a new rack. All of the parts are modular, and all parts can be sold back to Nvidia.
Nvidia does this because they have to. Without it, many customers would consider other vendors when upgrading, or delay purchases in case something better comes along later.
I would be very very curious in some source of these Nvidia buybacks
What does Nvidia do with the parts they buy back? Resell to others or grind to dust?
I doubt NVidia wants thriving second hand market, so I would guess grind to dust?
Forward them to their, partner Microsoft…
resell them? there is a decent market for people fine with less efficient, already used gear.
I'm not really sure they have thought this through. Or at least their solution has been to essentially insulate the company and even the physical building from the debt used to buy the GPUs. These SPVs typically lease the building and secure the debt to the GPUs themselves so if it all goes pear-shaped, it'll be the debtholders who are left holding the bag, priamrily.
But I have questions:
1. How often do GPUs fail? I've seen figures that this is as high as 9% annual failure rate for DC GPUs;
2. How easy is it to get replacement stock? At what cost?
3. If you trade in, what value do you get for your GPUs?
4. What is the availability like? How does NVidia decide who gets the next wave of GPUs? That's NVidia's big power here. It's a bit like how Boeing and Airbus decide on customer order and if you upset them, you may lose other orders. This happened to Qatar Airways over the A380 [1].
5. How long are the current form factors and interfaces going to last? How long before the power requirements get larger than what DCs are currently built for? Blackwell GPUs use a lot more power than, say, 2 or even 1 generation ago did.
GPUs are a huge cost factor, allegedly more than 50% of the cost of a data center [2]. If you take a bath on the trade in and are forced to do it because you can't compete with the current generation's performance-per-watt, that's not a good situation to be in.
[1]: https://www.arabnews.com/business/qatar-airways-in-mystery-a...
[2]: https://orennia.com/insights/what-it-actually-costs-to-build...
Sometimes I wonder if the people committing and spending this much money is really thinking about this or not?
My feeling is there is some kind of upgrade packaging done here. Like if the chips advance by X factor it can be offered at a massive discount or something? Otherwise it seems so silly to buy so much hardware which is evolving so fast.
I'm no fan of Oracle, but you could make the same claim about land developers, or datacenter operators, or REITs. Or a random conglomerate like Unileaver or GE or Siemens.
Ultimately, someone has to provide "commodities", and if you think your ship is the right vessel to do it betting hard on this is some kind of strategic direction.
Whether they are right, and to what extent debtors and the market will put up with cyclic and performance issues, etc.. I certainly won't hold the stock, but I wouldn't have earlier either.
This sounds a lot like the play that SpaceX made for their IPO.
ERPs are done. Data center permits, space, contracts to buy scarse GPUs and energy are the moats today. The chips can provide more tokers per watt but the world needs exponentially more tokens each day
Curious to know why you think ERPs are done. I actually do somewhat agree with you in that their best days are behind them.
It is going to be incredibly difficult to bill outrageous amounts for armies of consultant to implement or customise the product when agents can do the same for much less, or at-least a 5-10x reduction in headcount for similar complexity of changeset.
If anyone is in the industry, please feel free to chime in as to what you are seeing on the ground.
>Oracle does not have any proprietary IP that OpenAI depends on. That means Oracle is providing commodity services to OpenAI.
Huh? Is Oracle's plan with this funding to provide compute services for the big AI specialist companies? I assumed it was a buildout for their own AI services, as you're right, the former wouldn't make sense. (At risk of making a trollish, low-effort comment, OpenAI/Anthropic are savvy compute users and would have to be insane to risk getting entangled with Oracle.)
I read the article and I couldn't tell from it which of these is the case.
> their own AI services
Oracle doesn't have any. And that's wise. Oracle makes good RDBMS servers. They suck at everything else.
Their JVM doesn't suck.
On strictly business terms, you're absolutely correct. Considering that OpenAI and Oracle both eat from Trump's hand though, I think that Larry Ellison has a few tricks up his sleeve to remain a "necessary" contractor. This is probably a shorter-term investment to pad out the coffers 5-10 years from now.
If rampant despotism is the end-goal, Oracle has already aligned themselves with the despots.
A company that is purely in Denial, even doubling down at the risk of toppling the whole financial system should be placed under the financial equivalent of arrest and its leadership actually arrested. They're obviously playing with fire and it's alarming that no one's in the administration is batting an eye.
Oracle is small potatoes, it’s not toppling anything. If anything, Oracle’s demise would boost productivity.
Oracle not able to provide its usual "services" would disrupt many institutions which unfortunately rely on their technology and assistance operating the often expensive and complex Oracle integrations they've set up, for better or worse. I really despise Oracle and the Ellisons and they don't need a bailout but I'd hate to see what could come from many significant collapses of technology for social institutions.
If Oracle suffered a massive financial collapse wouldn't all those aspects of its business that are needed and paid for "just" get sold off and be managed as a new entity? I quote just because it would be an enormous amount of paperwork and organizational restructuring, but people with enterprise scale money tend to get to keep the services they pay for.
This came up in the last discussion: The TL;DR is that shareholders go to zero, bond holders take a haircut and then own "new Oracle" (Opacle?) that can generally sustain itself with revenues from agressive JDK audits, RDBMS licenses, and ERP/PeopleSoft.
(Opacle == One Poor A-hole Called Larry Ellison)
Half of that would be in stocks, resulting in 2.5% dilution for shareholders at the current market cap. But the stock is down 8%.
First homeopatic share.
It's not dilutive -- current shareholders get $10B (or 97% of $10B, whatever). Market reaction is some combination of profitability concern / bankruptcy risk because Oracle is super leveraged and taking on even more debt. Maybe the AI hyperscalars will slow down spending in 2028.
It’s literally dilutive. Current shareholders now only own 97.5% of the equity instead of the 100% they owned previously.
Yes, even if the company raises $10b, the ownership stake and voting power of a share is reduced by 2.5%.
And if that $10b figure is a fixed target, then if the share price continues to fall, the more shares they will need to issue, resulting in even more dilution.
https://en.wikipedia.org/wiki/Stock_dilution
Oracle begging couldn’t happen to a nicer company, the only thing that can be said is bubble, bubble bubble…
I would assume it's to hire more lawyers.
At this point it's a race between "how much money are left in the market" and "profitability". And both are hard to predict.
hopefully their demise brings JavaScript back to the commons!
https://javascript.tm
Previously, I thought investing in AI companies would be a good hedge against AI taking my job, but I hadn't considered the possibility that they fund the build out by issuing shares. Is there no reprieve?
The shares you own (usually) give you voting rights. Whether that makes any difference is another story.
But I'm not sure what people think the stock market is for? It exists primarily so that companies can access capital markets namely by issuing their own shares.
Historically for companies that were growing rapidly or had large capital needs, private markets weren't large enough to handle that. Times have changed, but the whole point of going public is to be able to raise capital by issuing shares.
maybe not. There could be mass unemployment and all the big dogs could also lose from the buildout while smaller players (non-public) take over. It could be a lycos / yahoo / google situation where there was no "winner" to invest in during the initial dotcom crash, and there was large-scale disruption all the same.
I'd like to note that "the sector that employs you" should be considered in diversification planning.
A stretch of unemployment can be weathered with good investments, or a crashing market can be waited-out by staying firmly employed, but both simultaneously is what forces people to "sell low".
The dotcom crash has nothing to do with the AI build out
Theoretically hedging would only work if the A.I. firms are not already priced for taking your job, right?
I have no opinion on whether they are.
I think they are actually underpriced relative to "taking away" jobs. I personally believe ant at 1-2T is pretty reasonable if you consider claude to effectively be the "best devtool of all time," aka it literally writes all software for you.
I think (somewhat arbitrarily), to get to higher valuations, you are going to have to disrupt more than just the software business. Which is what most labs are trying with finance, biology, chemistry, law, etc. That might be enough for a few hundred billion in TAM more.
I think to get to a ~5T+ valuation, then we have to start thinking about AI meaningfully displaces classes of jobs, maybe in the single digit %'s. As in, _global_ SWE headcount starts going down, but revenues for software companies as a whole go up.
I think for ~10T+, you definitely need meaningful job displacement across most white collar sectors.
Not for people like us.
It actually is but very few people realized that yet. We are still at "it doesn't work it's a fluke" denial phase
Broadcom is waiting to buy up the pieces.
Skynet finally? (Terminator movie used Oracle building as Skynet HQ, forgot which one)
We are b-roll in the skynet/oracle cinematic universe.
I'm happy to see Oracle missteps here.
I hope nothing but the worst for Oracle as a corporation...though nothing against the technical talent there.
Misread the title at first as "to finance AI bailout".
LOL.. It is that under the cover.
Of all the companies in the world this is one of the ones I root against the hardest. Hoping for the worst outcome on this investment.
Yeah I hope they get bought by Broadcom, like VMWare and the board sends Larry straight to retirement.
What a punishment, free up your days to spend your billions...
This is the equivalent of 2007 garbage CDSes.