FTC Launches Inquiry Into AI Deals by Tech Giants (nytimes.com) 12
The Federal Trade Commission launched an inquiry (non-paywalled link) on Thursday into the multibillion-dollar investments by Microsoft, Amazon and Google in the artificial intelligence start-ups OpenAI and Anthropic, broadening the regulator's efforts to corral the power the tech giants can have over A.I. The New York Times: These deals have allowed the big companies to form deep ties with their smaller rivals while dodging most government scrutiny. Microsoft has invested billions of dollars in OpenAI, the maker of ChatGPT, while Amazon and Google have each committed billions of dollars to Anthropic, another leading A.I. start-up.
Regulators have typically focused on bringing antitrust lawsuits against deals where the tech giants are buying rivals outright or using acquisitions to expand into new businesses, leading to increased prices and other harm, and have not regularly challenged stakes that the companies buy in start-ups. The F.T.C.'s inquiry will examine how these investment deals alter the competitive landscape and could inform any investigations by federal antitrust regulators into whether the deals have broken laws.
The F.T.C. said it would ask Microsoft, OpenAI, Amazon, Google and Anthropic to describe their influence over their partners and how they worked together to make decisions. It also said it would demand that they provide any internal documents that could shed light on the deals and their potential impact on competition.
Regulators have typically focused on bringing antitrust lawsuits against deals where the tech giants are buying rivals outright or using acquisitions to expand into new businesses, leading to increased prices and other harm, and have not regularly challenged stakes that the companies buy in start-ups. The F.T.C.'s inquiry will examine how these investment deals alter the competitive landscape and could inform any investigations by federal antitrust regulators into whether the deals have broken laws.
The F.T.C. said it would ask Microsoft, OpenAI, Amazon, Google and Anthropic to describe their influence over their partners and how they worked together to make decisions. It also said it would demand that they provide any internal documents that could shed light on the deals and their potential impact on competition.
The first question would be (Score:5, Insightful)
How come the only players OpenAI or Anthropic could have turned to to get enough computing power are Microsoft, Amazon or Google?
AI uses insane amount of CPU and data, both of which can be provided pretty much solely by the aforementioned Big Data oligopoly. By virtue of their very activity, OpenAI and Anthropic set themselves up to be bought or controlled by one of them, because without them, there's just no AI.
So the symbiotic and monopolistic nature of the deals between Microsoft / OpenAI and Google / Anthropic was inevitable. It's totally obvious that OpenAI and Anthropic are separate companies in name only.
The FTC should start by asking why it is that only the current incumbent players in the corporate surveillance economy could ever have provided the resources necessary to make AI happen, excluding virtually everybody else - even public funded academia - from the very beginning. Because typically preventing competition from even existing is a big red flag if you're looking for hallmarks of monopolistic behavior.
Re: (Score:1)
Ya get bigger by being big. Even Warren Buffett admits that he can make investments that are not viable to smaller investors because he can pool his risk such that no single failure takes the entire portfolio down.
Re:The first question would be (Score:5, Interesting)
How come the only players OpenAI or Anthropic could have turned to to get enough computing power are Microsoft, Amazon or Google?
AI uses insane amount of CPU and data, both of which can be provided pretty much solely by the aforementioned Big Data oligopoly.
The FTC should start by asking why it is that only the current incumbent players in the corporate surveillance economy could ever have provided the resources necessary to make AI happen, excluding virtually everybody else - even public funded academia - from the very beginning.
Do you recall the stories about how AWS came to be?
Amazon needed a lot of systems to handle their logistics, so they built datacenters and hired people to run them. They learned that they didn't need all of the capacity all of the time, so they tried leasing it out to other businesses when they didn't need it. This was a success. So then they added more capacity, because thanks to economies-of-scale it was profitable to run really massive datacenters selling capacity piecemeal to others. This business grew to eclipse the value of their original business.
Google and Microsoft followed similar paths: building massive datacenters to meet their own internal needs, and then building them out even further and leasing fractions of the capacity to others -in the process earning massive profits.
So, putting your paranoia aside, the answer to the question of "Why are the big cloud providers the best choice for providing lots of compute power?" is because they are the big cloud providers. Their other business divisions needs explain how they grew to be who they are, but their value is that they are who they are.
Why do we buy gasoline from the gas station? Because they are representative of the local oligopoly? Or because they sell what we need?
Re: (Score:2)
Exactly this.
OpenAI needed lots of A100 cards (which cost $15,000 in the used market alone), and Microsoft had a bunch of those in their Azure datacenters. (Google, AWS, Oracle, and many others also do). Why they chose Azure but not others? Who knows, maybe they negotiated better, or maybe they did not want to risk working with a competitor (Google/Deepmind), even though MS seems to have started competing with their in house AI efforts.
Anyway, not everything has to be a conspiracy.
As for the academia? It is
Re: (Score:2)
OpenAI needed lots of A100 cards (which cost $15,000 in the used market alone), and Microsoft had a bunch of those in their Azure datacenters.>
Pithier: there are about 10 billion reasons OpenAI would partner with a big cloud vendor.
Re: (Score:3)
Why they chose Azure but not others? Who knows, maybe they negotiated better, or maybe they did not want to risk working with a competitor (Google/Deepmind), even though MS seems to have started competing with their in house AI efforts.
ooh...I actually know this one.
Open AI started as a non-profit, with a promise of funding from Elon Musk. Due to a potential conflict of interest, Musk withdrew his support from the foundation, leaving Open AI in need of funding to pay for the massive amount of compute resources needed. Enter Microsoft.
Microsoft offered to give a Billion dollars in exchange for a license to use Open AIs yet-to-be-proven technology. But they didn't give a Billion in cash, but in Azure credit. Meaning Open AI had an almos
Re: (Score:2)
The FTC should start by asking why it is that only the current incumbent players in the corporate surveillance economy could ever have provided the resources necessary to make AI happen,
I'm pretty sure the FTC isn't really concerned with the "incumbent players in the corporate surveillance economy." Corporate surveillance has been going on for many decades, since well before Microsoft, Google, Amazon, and Apple were even conceived. And as for why these companies could "make AI happen"--it's kind of obvious. There is no way, for example, that "publicly-funded academia" could have or would have put up the money necessary to build the necessary AI infrastructure.
AI didn't just now appear. IBM
What they're gonna find will look a lot like (Score:3, Insightful)
AI is gross. It's either useless or fraudulent. The legal, productive use-cases for crypto do no represent the kind of scale that can be justified by the numbers that a tech monopolists are trying to convince us of. It's just a bunch of insanely wealthy scumbags creating another bubble.
Any company being investigated by the FTC, SEC or DOJ should be bared from govt contracts until they are fully exonerated.
Re: (Score:3)
You've clearly not actually tried to use AI.
Like you, I watched crypto burst upon the scene, and like you, I steered clear. I literally had a wannabe startup *beg* me to incorporate blockchain into my product, because blockchain was going to be the hot new thing. But for all their trying, they couldn't explain a single problem that blockchain could solve for my use cases.
AI, on the other hand, solves real problems every day. More precisely, it makes it quicker to solve problems every day. I never considered
glad to see it (Score:3)
I'm glad we have a FTC at this point that will look into things like this. That wasn't the case with the previous administration.
It's very possible that the major tech players will scoop up the companies doing the most significant AI and own the entire technology. 'Investing' the odd $10 billion here and there doesn't look as much like a purchase as actually buying the companies outright, but it may amount to much the same thing. Let's have a look.
to corral the power the tech giants (Score:1)
you need to demand the system be open and transparent (don't permit this nonsense [slashdot.org]), and limit patents and copyrights that conflict with that openness and transparency
OpenAI's recent fumble (Score:3)
I wonder if the FTC would be doing this if not for OpenAI's recent, and very public, fumble causing lots of chatter to be flowing around about the non-profit status being diametrically opposed to the seeming direction the company was taking? Perhaps that public spat from the board will finally get some form of regulatory eyes scanning these behemoth companies trying to corner the just beginning market on this type of AI?