Pop?
if it looks like a pop, swims like a pop, and quacks like a pop, maybe it is a successful business investment opportunity, not a pop
Close on the heels of a report that OpenAI has missed revenue targets and may not be able to pay its future bills, compute partner Oracle is keeping calm and carrying on with a massive new datacenter complex in the New Mexico desert. It seems like founder and chairman Larry Ellison is less concerned about whether Sam Altman …
This post has been deleted by its author
Leisure suit Larry owns billions of investments outside of Oracle. If Oracle went bust he'd still have something like a billion dollars a year to live off even if he lived too 150.
Oracle corp, they could easily go bust. Yet the name, all the assets, and half the employees would immediately be bought. Since Oracle is probably the software provider for 75% of US government ERP the new owner will be a big US financial corporation. Existing investors would lose everything, and a good chance the new owners would invite Larry back as their saviour.
Merely reporting as rough and ready facts, this DC alone would need almost the entire peak output (c2.9GW) of all grid scale solar farms currently operating in New Mexico. With solar in NM having a capacity factor of 27%, that wouldn't cover the other 73% of the time which needs storage or other generation. Rough calcs suggest that to cover the DC 24/7 needs from solar would need around 46,000 acres of solar panels, plus storage of an absolute minimum of 35 GWh of battery capacity, even that would only cover intra-day needs, not any longer periods of reduced sun.
Technically all pretty straightforward, but not cheap. Think $20bn as a ballpark figure for that scale of solar and storage. I'd also guess that the storage element costs go up in multiples if needed to cover for reduced solar events across the year.
China added over 250GW of solar last year.
2.9GW usage (steady 24/7) supplied from a 27% capacity factor represents about two weeks of Chinese solar deployment.
The question isn't whether solar can be built at that scale, but rather, why China does and we don't.
46000 acres is about 186 square kilometers, or a square 14km on a side.
Because Chinese solar panel manufacturers failed to make long term investments into US politicians in the way that oil producers did
The Chinese system of government ("party has to prop up failing manufacturing sectors") is fundamentally different to the American system ("political donations must be reported"), but no less corrupt. Payments were being made to local party officials, and "princes" were being cut in, when industry was rushing into the vast over-production of solar panels.
Then prices collapsed, and the party propped up the sector.
The question isn't whether solar can be built at that scale, but rather, why China does and we don't.
An easily answered question. Chinese deployment costs for any infrastructure are about a third or less of building in the US or Europe, and anything the Chinese government wants to happen just happens, regardless of payback or the sort of regulatory or legal obstacles that the West has.
For the most part yes. The over-build of Chinese infrastructure (ghost cities, unused international airports, high speed rail links or motorways carrying a trickle of traffic) would have caused a huge financial crisis anywhere else. In China they simple act as a state mal-investment that reduces how much it can spend on services or defence.
The "ghost cities" thing is mostly low-quality journalism.
Most of the "ghost cities" from 5 years ago are doing just fine now. It takes time for people to move in. Humans exhibit quite a bit of inertia in their housing decisions.
The real story here is that China has become so efficient in building housing stock that supply moves faster than demand. In the West, supply lags demand, even when housing is in shortage.
Major reason why the West has a problem with housing affordability. We're planning ourselves to death, strangled with a noose of red tape. The boomers and landlords are happy for housing to continue being in short supply.
Same with Chinese energy. They aggressively build out renewables, because they're efficient at producing them, which in turn prompts entrepreneurs to figure out how to make use of abundant power.
Adding housing to an existing city is a lot more difficult than finding a big open field somewhere and saying "let's build a city of 1 million right here" like China did.
The US would likely have similar "ghost cities" if we did that despite housing shortages, because there's a chicken and egg element to building a new city. Who is going to want to be the first people to move into an empty city? Who are the first companies to move into an empty city that has no employees for them to hire or any customers?
Part of the reason it has worked at all in China is because the US model is "build it, then sell it" while China's model is "sell it, then build it". Here in the US I don't have to pay a cent until the house I buy is complete and ready to move in tomorrow (you can buy it before it is complete, if you want it more customized, but you that's not the way most new homes are sold)
You'd almost need to have some sort of government subsidy to encourage people/companies to move there - maybe offer them no taxes until the city is 50% occupied or something like that.
> "The US would likely have similar "ghost cities" if we did that despite housing shortages, because there's a chicken and egg element to building a new city. Who is going to want to be the first people to move into an empty city? Who are the first companies to move into an empty city that has no employees for them to hire or any customers?"
China's system is efficient at solving coordination problems, by design. Western systems (particularly the US) are not efficient at solving coordination problems, also by design.
Who can move in immediately? Remote workers, retirees, etc. This would relieve housing supply pressures elsewhere. Once the settler class moves in, the rest of the economy follows. We should be giving a lot more thought to housing remote workers and retirees, in cities built with infrastructure as a coordinated design priority. It benefits everyone.
China did things a bit differently (before RW and an uptick in retirees): industrialization of agriculture caused huge demand for cities among former farm workers who wanted to leave rural areas.
> "You'd almost need to have some sort of government subsidy to encourage people/companies to move there - maybe offer them no taxes until the city is 50% occupied or something like that."
Considering the cost of subsidies and poorly-scaled efforts to make unaffordable areas "affordable", it's an open question whether that might be both cheaper and more effective than current efforts, which have mostly failed (and at high cost).
The West has become allergic to building stuff. Not just housing.
"Who are the first companies to move into an empty city that has no employees for them to hire or any customers?"
It's not a good idea to plop down a whole city far from any others, but large companies can take some time to outfit/build a new facility which means there's time to recruit people. If there's housing and utilities in place, it's easy for those companies to get up and running more quickly than if they locate to an established city with housing limitations. When housing is in short supply, sellers and landlords are going to leverage their position to get top dollar. If there is more supply, homes can be available at more reasonable prices.
Most of the "ghost cities" from 5 years ago are doing just fine now.
"S&P notes that secondary home prices have declined for 44 consecutive months, and rents have fallen for 23 months."
Vacancy levels are still up 45%.
Most analysts (not just "low quality") are now looking for an end to the 5 year housing collapse. The Chinese have announced a 5-year recovery plan. So expect the sector to be doing "just fine" 5 years from now.
Much better that housing prices keep going up to infinity, right? Much better to be in the US, where a plurality of first time homeowners inherit a house from parents, right?
China has prioritized lowering the cost of life essentials, like housing.
Trump prioritized an idiotic war in Iran which jacked up energy prices and spread more inflation through an already inflation-ravaged economy.
Most analysts (not just "low quality") are now looking for an end to the 5 year housing collapse. The Chinese have announced a 5-year recovery plan. So expect the sector to be doing "just fine" 5 years from now.
Hahahahhaaa!
https://www.ft.com/content/bbef9296-2e69-4185-b7f1-f0c8fa01030c
"Chinese deployment costs for any infrastructure are about a third or less of building in the US or Europe"
It seems to me that a lot of that is streamlining after the decision has been made to allow a project to go forward. In the US, the permit/inspection/Fee based systems that operate at all levels of government do the most to slow down and add costs to projects. Time after time on the house flipper shows, the developers would have to wait a month or more to get a design approved or an inspector to come out and sign off on something that's been done. That plays hell with scheduling as crews and trades bugger off to other jobs during those delays and it takes time to get them rescheduled or replaced. When Tesla built their plant in Shanghai, the inspectors were onsite nearly continuously and checking that things were done correctly in real time. No waiting. Approved plans in hand and the project could go full steam with scheduling much easier to manage.
Did they really, or did they just say they did? I do not trust the Chinese government's word as saving face is far more important than anything else in Chinese culture. Nothing wrong with saving face on the face of it (ahem) but when coupled with a totalitarian government, nope.
"Technically all pretty straightforward, but not cheap. Think $20bn as a ballpark figure for that scale of solar and storage."
Whether Oracle spends that money or not, somebody has to spend the money to provide the power. Most of NM seems to be the White Sands Missile base or other government installations with a very high poverty rate surrounding those places. It's not known for having abundant water nor much land suitable for farming. It might be a good plan to cover a lot of it in solar panels and shift power hungry industries to the state (or make it easy for those companies to make the move). Slowly at first with an eye towards how those solar farms impact the local climate with expansion if it doesn't lead to ruining farms in the middle breadbasket of the US. Energy storage is going to be an issue for when the sun doesn't shine, but that's easier to manage when there is plenty when the sun is shining.
With a large solar infrastructure, electrifying rail and having plenty of EV/ET charging stations would be easy. I'm not a fan of HSR in the US until there is more passenger rail services at 125mph, but at some point, an express HSR line going East and West would make good sense to route where much of it could be powered by solar and maybe wind.
The size of the US (and hence long distance travel needs) means that for high speed rail to be a credible solution 125 mph is too slow. For a country like Britain where even our longest journeys are shorter between stops, 125-140 mph are about as high as is feasible when the practicalities and economics are factored in, but in the US those speeds would only be suitable for in-state travel.
"The size of the US (and hence long distance travel needs) means that for high speed rail to be a credible solution 125 mph is too slow. "
For travel from coast to coast or North to South, yes, it would be better to have HSR, but at that point, flying will be much less expensive and lower cost. A train that can average 100mph can cover 1,000 miles from 8pm to 6am. That's a reasonable time for a proper sleep and close to half way across the US, E<->W. Doubling the speed will mean leaving at 8pm and arriving at 1am, which is not that useful. It will also be expensive to lay down 1,000 miles of grade separated track. The amount of money for all of that only benefits a small number of travelers where the same money spent on higher speed service and adding trains to the schedule will service more people and give more option. Some long distance trains in the US only operate 3 days a week and many stops will be at small stations in the wee hours of the morning when there are no transportation options or lodging open to guest arrivals. Some routes have exceptional scenery or would if the train went past during daylight hours. I'm fine with leaving late, but I don't want to arrive in the middle of the night.
50mph can be a good speed for some trains. Leaving Los Angeles in the evening and going slow to arrive near San Fransisco first thing in the morning or London to Edinburgh overnight can be an excellent way to save days on a holiday. It could also mean only spending an effective day at the destination and getting back ready to the next business day. I always see flying as taking a whole day with all of the strip searches and other security kerfuffle. I could do a round trip between two cities with the air portion of travel taking around an hour, but I'd be half the day answering why I booked the day before rather than 3 weeks in advance and might miss the flight if I don't build that extra time into when I arrive at the airport, both ways. Such are the realities of business travel.
I concur that NM would be a good candidate for "covering with solar", but as to the concern of it having impact on the breadbasket, I'd suggest it might - albeit in a slightly different way to that as expected.
I cite the outstanding recent Technology Connections video where he cited (with accompanying back of a cigarette packet math) that the quantity of corn grown in that part of the US for bio-ethanol for gasoline is an order of magnitude less efficient than an equivalent area in an equivalent location covered with solar panels to run an equivalent fleet of EVs. Such was the argument that even with obscene allowances for the figures in favour of the corn, it's still a compelling argument. "Covering NM with solar" and an EV push would significantly reduce corn-for-fuel reliance and thus the saleability of the corn, or force a plan for secondary markets.
Back to the point for Oracle, I understand the shenanigans that make "fuel cell" the preferred approach and scare from sufficient battery storage, but that doesn't negate my disappointment that they're not choosing (or forced) to install sufficient solar to operate independently during peak daylight, or to cover the site in-full if sufficient space doesn't exist, and lower the requirement on actual non-renewables.
Roughly 40% of US corn production is used to manufacture ethanol fuel. That number is expected to go up, now that the Trump administration has approved nationwide year-round sales of E15.
The land efficiency is indeed terrible from an energy perspective. Solar + battery does a much better job of capturing and storing primary energy than plants, and with far less associated environmental impact from things like fertilizer and pesticide use.
Ethanol is one of the worst boondoggles in Washington and keeps expanding due to the influence of special interests. It's caused a humanitarian disaster in the third world by driving up food prices and soaking up supply. Domestically, ethanol competes with animal agriculture for feed corn, severely driving up the price of meat. The corn monoculture has created biosecurity risks. The farmers themselves also have a certain corn dependency, with many lacking extensive knowledge of other crops. Systemically, our markets (particularly export) for other crops could use more development.
The energy transition will be bringing the era of corn ethanol to an end, and probably sooner than forecast now that the Iran war has many more Americans questioning how long they personally want to stick with liquid fuels. If the transition is disorderly, it's going to be utterly devastating to farming communities. Funding for agriculture extension resources and crop research has been slashed, the tariff wars have alienated foreign markets, and there's no coordinated effort to begin planning for the post-ethanol farm economy.
Sad, really, because this is some of the world's most productive farmland, and US farmers have competitive advantages in access to capital, access to technology, and overall scale. If only that land could be used to feed people. Plenty of land available for solar where farm production is nonexistent, where efficiency is marginal, or where farms are living on borrowed time due to depletion of fossil water resources.
"I concur that NM would be a good candidate for "covering with solar", but as to the concern of it having impact on the breadbasket, I'd suggest it might - albeit in a slightly different way to that as expected."
I watched Alec's episode on that as well and it makes good sense. A big change in albedo over a large area can affect weather patterns and NM is at the southern border of a lot of the US mid-west growing belt. China has been doing some experiments with planting solar farms and some agriculture on the edge of one of their big deserts to hold back desertification. They are seeing the effects and they are mostly good, but that there are observable effects means there must be caution since every situation will be unique and changing weather patterns can be unpredictable.
Putting solar farms in the corn belt and stopping growing food to turn into fuel can be a double win. Farms need power and if there is land on farms that isn't useful for growing, sun becomes another crop they can sell if it's close enough to a customer (electric cars/trucks?).
I would have thought that there is plenty of sunshine in New Mexico.
Not at night. So there's a need to invest in storage. Someone current on costs should pencil out what they might be.
Bloom's primarily run on natural gas. They can be converted to run on hydrogen, but for that to work, you also have to find a supply of H2 that was sustainably sourced, which is harder than it sounds.
There's a possibility that direct solar production of hydrogen might become more efficient. At which point, the Bloom investment isn't lost.
> "Not at night. So there's a need to invest in storage. Someone current on costs should pencil out what they might be."
Wall Street already has.
Clean energy wins on the numbers.
Start on page 8.
"There's a possibility that direct solar production of hydrogen might become more efficient."
You need to post a reference for that. Electrolyzing water is dominated by the energy required to break the molecular bonds. Nobody has yet come up with a 100% efficient way to do that in bulk where all of the energy going in is laser focused on the bonds. There are plenty of people that don't understand the basic energy requirements and keep on believing that "science will find a way". When up against physical laws, there isn't any wiggle room.
"Oracle is pitching it as a green move nonetheless"
Well, it most certainly isn't a green move as this datacentre will be powered by a fossil fuel. If Larry had agreed for this datacentre to be powered by a new solar farm in sunny New Mexico then he would have a valid point but he doesn't so it's all just total BS on his part.
Fortunately we have fiber optics to move data to bit barns which can be located far enough away from any people they might bother.
The US has plenty of open land. Trouble is much is unavailable for development. They can't impact the habitat of orange-striped tree weevils in a remote area, so they'll just put it next to your neighborhood and bother you instead.
What's obvious to any 12 year old who has ever played a city builder game is, unfortunately, not obvious to the politicians and planning "experts". Or it is, but special interests often have other priorities.
Natural Gas, oil, and coal all have the same carbon footprint.
When burned, they release the same amount of CO2 per unit of energy.
He's calling it a fuel cell to avoid saying fossil fuel power plant.
The government deemed CO2 the primary concern that will cause floods, droughts, hotter summers, colder winters and the end of the earth.
As a result, our access to the cheapest and most abundant energy source has been restricted through taxes and legislation.
Why doesn't this apply to billionaires? The energy they create will not heat homes. It will be used to replace employees and make anime porn.
This should concern everyone
Oracle was always a software company. They got some hardware chops by purchasing Sun Microsystems but its unlikely that many Sun people are still around. This is a view of a software company building data centers in 2026. Oracle''s SAAS offerings are safe. Its not clear if Oracle will be long term in the public cloud business. Without Larry's grandiosity, probably not.
Natural Gas, oil, and coal all have the same carbon footprint.
When burned, they release the same amount of CO2 per unit of energy.
He's calling it a fuel cell to avoid saying fossil fuel power plant.
The government deemed CO2 the primary concern that will cause floods, droughts, hotter summers, colder winters and the end of the earth.
As a result, our access to the cheapest and most abundant energy source has been restricted through taxes and legislation.
Why doesn't this apply to billionaires? The energy they create will not heat homes. It will be used to replace employees and make anime porn.
This should concern everyone