Nebius Group NV
Nebius Group N.V., operates as a technology company that engages in building full-stack infrastructure to service the global AI industry. Its businesses include Nebius, an AI-centric cloud platform built for intensive AI workloads.
Bloom Energy Corp
Bloom Energy is an American public company headquartered in San Jose, California. It manufactures and markets solid oxide fuel cells that produce electricity on-site. Bloom Energy (BE) Gains “Buy” Rating as Data Centers Boost Power Demand
Vertiv Holdings Co
Vertiv is an American multinational provider of critical infrastructure and services for data centers, communication networks, and commercial and industrial environments.
Vistra Corp
Vistra Corp. is a Fortune 500 integrated retail electricity and power generation company based in Irving, Texas. The company is the largest competitive power generator in the U.S. with a capacity of approximately 39GW powered by a diverse portfolio, including natural gas, nuclear, solar, and battery energy storage facilities.
How Vistra Corp. (VST) is Positioning Itself for AI-Driven Electricity Demand with Nuclear Power
Nuscale Power Corp
NuScale Power Corporation is a publicly traded American company that designs and markets small modular reactors. It is headquartered in Tigard, Oregon. A 50 MWe version of the design was certified by the US Nuclear Regulatory Commission in January 2023.
Modine Manufacturing Co
Modine Manufacturing is a thermal management company established in 1916 in the United States. The company started as Modine Manufacturing Company by Arthur B Modine who patented the Spirex radiator for tractors. The Modine company manufactured the Turbotube radiator for Ford Model T cars.
Modine’s data center sales soared 102% year over year in fiscal 2025 Q2
We are currently focused on supporting trends like high-performance computing as the rapid expansion of AI and machine learning fuels demand for data center capacity and more advanced cooling solutions,” CEO Neil Brinker said on the call. “Although these trends can change over time, we believe that the need to meet ever increasing regulations and reduce the impact of fossil fuels on our environment [is] currently driving multi-year growth cycles.”
Credo Technology Group Holding Ltd
Credo’s mission is to deliver high-speed solutions to break bandwidth barriers on every wired connection in the data infrastructure market. We provide innovative, secure, high-speed connectivity solutions that deliver improved power efficiency as data rates and corresponding bandwidth requirements increase exponentially throughout the data infrastructure market.
Oklo Inc.
Oklo Inc. is an advanced nuclear technology company based in Santa Clara, California. Founded in 2013 by Jacob DeWitte and Caroline Cochran, both graduates of the Massachusetts Institute of Technology, the company designs compact fast reactors with the aim of providing clean, safe, and affordable energy. Its chairman is OpenAI co-founder Sam Altman.
Oklo Inc.’s (OKLO) Role in AI Infrastructure: A Buy Signal?
IREN Ltd
We are the responsible way to power Bitcoin, AI and beyond. We own and operate next-generation data centers powered by 100% renewable energy. We target regions where there are low-cost, abundant and attractive renewable energy sources, and where we can help solve energy market problems.
Iren CEO on powering Bitcoin and AI and scaling data centers
Coherent Corp
Coherent Corp. is an American manufacturer of optical materials and semiconductors. As of 2023, the company had 26,622 employees. Their stock is listed at the New York Stock Exchange under the ticker symbol COHR. In 2022, II-VI acquired laser manufacturer Coherent, Inc., and adopted its name.
Talen Energy Corp
Talen Energy is an independent power producer founded in 2015. It was formed when the competitive power generation business of PPL Corporation was spun off and immediately combined with competitive generation businesses owned by private equity firm Riverstone Holdings. Following these transactions, PPL shareholders owned 65% of Talen’s common stock and affiliates of Riverstone owned 35%. As a condition for FERC approval Talen agreed to divest approximately 1,300 megawatts of generating assets in the PJM Interconnection Region to mitigate FERC’s competitiveness concerns. On December 6, 2016, Riverstone Holdings completed the purchase of the remaining 65% of Talen’s common stock, making it a privately owned company.
Buy 5 Nuclear Energy Stocks to Tap Huge AI-Powered Data Center Growth
Semtech Corp
Semtech Corporation is an American supplier of analog and mixed-signal semiconductors and advanced algorithms for consumer, enterprise computing, communications and industrial end-markets. It is based in Camarillo, Ventura County, Southern California. It was founded in 1960 in Newbury Park, California. It has 32 locations in 15 countries in North America, Europe, and Asia. Semtech is the developer of LoRa, a long-range networking initiative for the Internet of Things. As of March 2021, over 178 million devices use LoRa worldwide. LoRa has been used in satellites, tracking of animals, UAV radio control, and natural disaster prediction, Semtech has been publicly traded since 1967. In 1995, Semtech ranked fifth on the Bloomberg 100 list of top-performing stocks of 1995 on the New York and American stock exchanges and the NASDAQ National Market.
Come on go lower.Every lake on Earth juat breathed a tiny cautious sight of relief
turning tech stocks red
oh is this just hitting GPU manufacturing (NVIDIA, ASML) and companies that were named in Trump’s AI thing? lol. Google being down 1-2% is a normal day.
TSMC is already into -10% territory, this is dragging a lot of different AI adjacent stocks down now
yeah tsmc is part of the gpu supply chain so its not surprising theyre being hit if asml is being hit.
Down 14% now
idk this plus the tariff thing is gonna royally fuck the US economy
No. I’ve updated my OP, its power generation, its datacenter operations, datacenter construction, data center cooling, its everything connected to the operation of a massive AI datacenter.
Line
Go
DOWN
deleted by creator
I don’t follow finance but doesn’t this prove that AI is just a bubble?
It proves that large sums of cash are not what predict powerful outcomes regarding the AI market. Deepseek was trained using only $5mil and without access to NVIDA’s top-tier hardware. It performs just as well as OpenAI’s latest model at a fraction of the resources required. Coupled with the R1 model being released for free under an MIT Open-Source license and the methodology to produce these results published publicly, anyone with the resources can replicate this method, rendering any proprietary methodologies obsolete. Older AI hardware from NVIDA that might not have run OpenAI’s latest model, could very likely run the Deepseek R1 model, leaving NVIDA holding the bag. This comes on the heels of Trump announcing a joint venture investing up to $500 billion for infrastructure tied to artificial intelligence by a new partnership formed by OpenAI, Oracle and SoftBank. Now it would seem that $500 billion could be overkill.
Edit: I’ve updated my post, and HOLY SHIT. It’s not just NVIDA, AMD, Open AI. It’s cascading across the entire tech sector and all the way out to Power Generation corporations. From Nuclear Power, to Green Energy, to Datacenter Maintainers, Datacenter builders, Domestic Semiconductor producers, fucking everything.
We’re talking probably TRILLIONS of dollars wiped clean off the market.
just to add: the cost of training DeepSeek was less than the yearly salary of one manager for most of these other projects, and they had 20+ at a lot of these big western AI firms
also before leaving Biden placed tiered restrictions on the number of newer AI GPUs that Nvidia can sell to enemy countries, and some ‘allies’ like Portugal or Poland also got placed in the ‘middle category’ which was 50k max or 100k if you sign some agreement (Polish commentators lamenting ‘but we thought we were the closest allies’ was very fun to see), but I guess now that restriction doesn’t matter as much because it looks like you don’t even need the fancy ones to get comparable chatbot results so that’s another
bucket of cold water on the US
they forgot PolandNot disagreeing, to add, ‘AI GPUs’ are a crank term created by Silicon Valley and their lobbyists to jack up the price on newer western GPUs in the west and restrict access to China. TSMC doesn’t recognize it as an industry term, there are no industry studies that show any kind of significant performance difference for LLM work, as now additionally proven by the DeepSink model.
Our world is predicated on bullshit and shareholder feelings
anyone with the resources can replicate this method, rendering any proprietary methodologies obsolete.
It’s not just that it’s easily replicable elsewhere. It basically destroys any hope for these companies to be profitable (at least with this stage of development). They’ve sunk too much, they can never hope to compete AND make a return on the initial investment.
Oh I totally agree. Maybe this will wipe all these idiotic AI initiatives off the map, and we’ll get a leaner, more interesting market for this technology that also isn’t sucking the environment dry.
that’s neat!
-$$% 
Basically, yeah. It definitely proves that Western tech companies were spending several times more money to make their LLMs than was truly necessary, and it means that they now have to try to catch up to DeepSeek if they want to be competitive (they actually have to find ways to exceed DeepSeek significantly since DeepSeek is open source, no one is gonna pay thousands of dollars for a corporate LLM that’s only 10% better than the free one).
I agree with the majority of your comment.
no one is gonna pay thousands of dollars for a corporate LLM that’s only 10% better than the free one.
This is simply not true in how businesses actually work. It certainly limits your customer base organically but there are plenty of businesses who in “tech terms” overpay for things that are even free because of things like liability and corruption. Enterprise sales is completely perverse in its logic and economics. In fact most open source giants (e.g. Redhat) exist because of the fact that corps do in-fact overpay for free things for various reasons.
Also anyone who works in tech knows that plenty of tech company owners and boards get incredibly caught up in hype and will absolutely make dumbass financial decisions to purchase products on hype alone. It’s ridiculous but it happens all the time. When it busts they just lay folks off and blame shit like “we expanded too fast”
Haha… Boy do I have stories… I worked in a terrible evil company (aren’t they all but this one was a bit egregious).
The CEO was an absolute moron whose only skill was being a contracts guy and being a money raising guy. We had an internal app for employees to do their work on in the field. He was adamant about getting it in the app store after he took some meeting with another moron. We kept telling him there’s no point, and there’s a shit ton of work because weh ave to get the app to apples standards. He wouldn’t take no for an answer. So we allocated the resources to go ahead, some other projects got pushed way back for this.
A month goes by and we have another meeting, and he says why isn’t X done. We told him, we had to deprioritize X to get the app in the app store. He says well who decided that. We tell him that he did. You know how a normal person would be a bit ashamed of this right? Well guess what he just had a little tantrum and still blamed everyone else but himself.
Same guy fired a dude (VP level) because his nepo hire had it out for him. That dude documented all his work out in the open, and then when that section of the business collapsed a day later they had to hire him back as a contractor and the CEO still didn’t trust him and trusted his nepo hire, and didn’t see the fact that his decision making was the inefficiency.
When I retire I swear to god I’m going to write “this is how capitalism actually works” books about my experiences working with these people.
When I retire
Damn you are gonna retire? I’m 40 and already resigned myself to working till I die.
By “retire” I mean, when I have aged out of software and I can just burn all my bridges.
Yeah, I have an oestensibly white collar job (although that hasn’t really meant anything for about two decades) and same.
I’ve worked retail for about 20 years and now it looks like my only way out is total economic collapse.
Yep this is basically exactly how every tech company is run
Throwback to that company putting “blockchain” in its name and having a massive increase in valuation despite no operational changes occuring
IBM named all their Field Support Technicians (previously: helpdesk) peons “AI Analyst”. It’s all a game for valuation.
Sums up my last job perfectly
Yeah I guess you’re right but I think even taking that into account, I’m confident a lot of startups will spring out of the ground that will be developing DeepSeek wrappers and offering the same service as your OpenAIs at a fraction of the price. So it’s still trouble for Silicon Valley AI companies.
I’m confident a lot of startups will spring out of the ground that will be developing DeepSeek wrappers and offering the same service as your OpenAIs
This is true. But I don’t think OpenAI is even cornering the tech market really. The company I work for makes a lot of content for various things and a lot of engineers are tech fetishists and a lot of executives are IP protectionist obsessives. We are banned from using publicly available AI offerings, we don’t contract with Open AI but we do contract with Maia for creating models (because their offering specifically talks through the “steal your IP” problems). So OpenAI itself is not actually in many of these spaces.
But yeah your average chat girlfriend startup is going to remove the ChatGPT albatross from its neck, given it’s engineers/founders are just headlines guys. A lot of this ecosystem is really the “Uber but for <MY THING HERE>” style guys.
At least on the hardware side, a lot of things boil down to businesses needing a designated target to push responsibility to whenever shit hits the fan. Ultimately, Redhat et al get paid because they’re willing to be the designated target whenever some dumbass IT manager fucks up and doesn’t want to take the heat. That’s why they’ll toss 3 year old workstations even though those workstations just spend the last 3 years intermediately running Outlook and Teams. They do this because 3 years is usually how long the warranty lasts. But once the warranty is up, they would rather buy new machines and the warranty that comes with it than support those practically new machines.
This is incidentally a great way to get an almost brand new PC at bargain prices. Just find whatever Optiplex or equivalent model that was released 3-4 years ago and buy a refurbished one.
The market can be wacky and they could recover. Looking pretty bubbly though.
woah hey explain what the hell is happening. what is deepseek what does this mean.
DeepSeek is an AI model released out of China, that seems to perform on par or better than GPT 4o. But this one was a side project of some dudes, and the whole thing cost something like $6mil and uses older, less “advanced” hardware. In comparison, the OpenAIs and Metas in the west have sunk untold billions into their project, and have no hope to compete in profitability. It’s glorious
It really is a Sputnik moment for the West, except this time the communists are the ones with the much stronger industrial base(d). Shit’s wild
It’s insane to see how much more optimized this is. I don’t know how I feel about it being open source though, USians can just claim it as their own
I’m sure there are more knowledgeable people on this than me here, but I was under the impression that they’ve open sourced the weighting and the general approach. Have they also released their training set?
In any case, I’m not too worried. The reason OpenAI, Meta etc. got to where we are are structural - the way VC and the stock market work, the way these soulless corporation politic and hoard resources, the way the tech hype cycle flows are why we got here. Getting to copy the Chinese kid’s math homework ain’t gonna change any of that.
On the contrary.
The business mode of all the American AI firms is to sell subscription to peoples and businesses to use their most advanced AIs, so giving everyone access to a model of similar performances for free destroys that business model.
Se even if American firms did steal it, so what? They’re not gonna make much money from it if peoples and companies can just download the free open source one instead of buying theirs.
Somewhat, but from anecdotal experience with coworkers, there are government contractors out there that exist solely to provide support for an open source product.
An example told to me was that products needed support from a vendor, so they would contract out support for an open source product, which would also contain a “guarantee” that the source code had no malware in it.
A load of bologna but people still somehow managed to monetize a free product
Moore’s law btfo
i’m barely informed about what any of this actually means but china released an AI model that apparently uses 50x less energy than chat GPT and is just as good. apparently it’s also open source.
i don’t really know what all this implies in reality, if it means all of these drops we’re seeing in the stock will stick and these AI billionaires in the US will lose some of their power in the longrun, or what.
The open source part is the biggest thing imo. Any company that has been paying OpenAI hundreds or thousands of dollars a month to utilize ChatGPT for their website, customer service, or for each employee to use individually, can now just implement DeepSeek for “free” (not including cost of implementation and energy to run)
My China ETFs and mutual funds have been going up while S&P500 is starting to tank.
I should have bought some short ETFs on Friday as they are going through the roof.
Wait are there some china etfs you can recommend? I’ve been trying to invest in their success for a long time
death to
Not financial advice, just my observation. Most of them seem to do well. I use Fidelity, and I have bought shares from KraneShares (such as KALL), iShares (such as FXI), and Fidelity (such as FHKCX (mutual fund)). KraneShares and iShares seem to do better than Fidelity ETFs/mutual funds. I also invest in FXAIX and IVV, which are Fidelity and iShares’s equivalent of S&P 500. As bubbles start to burst and Western markets crash, I plan to switch more of my US investments to China and emerging market ETFs.
I have lost some money by choosing my own stocks in the past and waiting too long to sell. I have made money back by switching to ETFs and mutual funds.
Thanks comrade
Deep Seks initiated
But at what cost?!
Sorry old friends, but my retirement account is going to love the rebound, however long it takes. The portion invested in Chinese yuan will make up for it otherwise.
did you just buy yuan or buy some sort of financial instrument to bet on yuan
I was half memeing. I just have my account overweighted (compared to common advice) in international/“foreign” equities versus domestic because I expect countries like China to be dominant by the time I actually retire.
Investing in the Chinese stock market isn’t usually profitable though!
In the short-term I invested in Chinese ETFs last year, I had 20-45% returns.
- -$$%
If I would want to " buy the dip" which of these capitalist nightmare makers would be most prudent to throw money after?
Maybe Nvidia since they make actual material goods. Don’t take my word for it, though, I’m just guessing lol
Good thing I didn’t lmao
