When you use an AI chatbot, such as ChatGPT from OpenAI or Bard from Google, you’re actually interacting with the result of three or four main ingredients.
One is the technical skill required to create the AI model for the chatbot. Another is how much training data the model ingested in order to become proficient at responding to your commands. Thirdly, even with the quickest semiconductor chips, the training process can take months due to the sophisticated semiconductor chips used.
Cloud platforms are now quickly rising to the position of the fourth necessary component in AI. They rent out all of that computing power—which they aggregate from many of those in-demand semiconductor chips—as well as online storage and other services to AI startups that require raw processing power and a location to store their training data.
The ebb and flow of the AI business is shaped by the reliance of AI developers on cloud services, placing cloud companies at the center of a technology that has the potential to revolutionize work, play, and learn.
The half-trillion dollar cloud business is dominated by a small group of very large companies, namely Google, Microsoft, and Amazon. Politicians and industry watchdogs now caution that Big Tech could have a huge and potentially anticompetitive effect over AI in the future due to its dominance in cloud markets.
Sen. Elizabeth Warren, a Democrat from Massachusetts, expressed her extreme concern that a small number of Big Tech companies control cloud computing and storage. These businesses will squash competitors, solidify their control over AI, jeopardize innovation, consumer privacy and safety, and national security in the absence of sensible legislation. In this vital sector, we have to safeguard competition.
For this report, all three corporations declined to comment. Proponents of the industry contend that the top businesses in the field fight hard for clients and that their size enables AI firms to find a one-stop shop for all of their cloud computing requirements.
However, policymakers’ concerns focus on the possibility of inflated pricing, anticompetitive collusion, exploitative contract terms, or other practices that may dictate what it ultimately costs to use AI services or the kinds of products that make it to market—just like with any other crucial industry controlled by large players.
There are only a few significant participants in the cloud business.
As generative AI gains momentum, AI’s percentage of overall cloud spending is anticipated to increase significantly.
Artificial intelligence may account for between 30% and 50% of the public cloud market in as little as five years, according to Gartner and other industry analysts, despite the fact that the overall market is growing at an exponential rate and that spending may increase by more than 20% to $679 billion next year.
According to Matthew Prince, CEO of internet monitoring and security firm Cloudflare, only a select few cloud platforms are able to provide the massive processing volumes that AI firms and their clients are becoming more and more in need of.
According to Prince, cloud providers will undoubtedly play a larger role in determining the winners and losers of the artificial intelligence market as AI developers and other firms seeking to adopt AI grow more reliant on them.
According to market research firm Gartner, artificial intelligence currently represents a very modest percentage of the $563 billion public cloud industry—possibly less than 10%.
Largely speaking, the top three cloud suppliers of AI are Amazon, Microsoft, and Google. Smaller cloud providers like IBM and Oracle are ranked after them.
Stricter regulations both domestically and internationally
Governments everywhere are observing the developments in a sector that is becoming more and more well-known.
President Joe Biden and the Federal Trade Commission in the US have both raised concerns regarding cloud market competitiveness as a possible roadblock to AI advancement.
The FTC stated in a June letter that, as AI is becoming a more integral aspect of everyday life, a few corporations that control a large portion of the cloud industry or any other crucial area of AI development might have a disproportionate amount of power over a large portion of the economy.
In October, Biden addressed artificial intelligence with an executive order that he signed, hinting at such worries as well.
The judgment stated that fostering an equitable and transparent market for AI development necessitates putting an end to illegal cooperation and addressing the dangers associated with dominant corporations using crucial resources like data, semiconductors, processing power, and cloud storage to disadvantage rivals.
The directive made no mention of particular suppliers or any wrongdoing on the part of the sector.
The FTC and other international competition authorities have conducted several in-depth investigations into the cloud industry in recent years, closely examining the market.
For instance, a UK government examine this fall focused on Microsoft and its software licensing policies, which have little to do with artificial intelligence. In March, the Federal Trade Commission (FTC) said that it was investigating the sector in conjunction with France, Japan, the Netherlands, and South Korea.
This is a highly concentrated market, according to the majority of regulatory inquiries to date, according to Sarah Myers West, managing director of the AI Now Institute and a former senior advisor on AI for the FTC.
Cloud platforms collaborate with AI startups in part because, as West noted, if their solutions are used by businesses or consumers, there is an ongoing demand for additional cloud resources.
Additionally, it can set up big tech corporations to gather even more information from internet users and businesses.
The simplest way to benefit from these technologies is for them to “walk people into their cloud ecosystems,” the source stated.
What cloud critics say they worry about
Some industry observers have identified a few specific cloud practices that could give rise to worries regarding future competition.
One example are the sometimes exclusive contracts that AI businesses sign with one or two cloud providers in return for substantial sums of money.
According to West, artificial intelligence developers typically receive exclusive partnerships for cloud computing, in contrast to other industries where it may be more common to use various cloud providers.
According to her, several of the AI firms developing solutions for the market today have already signed exclusive licensing contracts with cloud providers. Even a few of the smaller players mostly use that one [provider] and have ties to a cloud infrastructure company.
Consider the case of OpenAI. The ChatGPT tool developed by the firm, which last year astonished customers and ignited the current AI frenzy, is largely dependent on a $13 billion cash and free cloud computing usage contract with Microsoft.
In November, Microsoft CEO Satya Nadella told that OpenAI would not exist without Microsoft fully supporting the company’s goals. Nadella also said that due to the recent governance crisis at OpenAI, Microsoft could do all of the startup’s AI work on its own if it had to. They have employees, computers, data, and everything else they need.
Amazon stated earlier this year that it would invest up to $4 billion in the AI startup Anthropic in return for a small stake and the title of “primary” cloud provider. A few months ago, Anthropic made a deal with Google to be its “preferred” cloud service.
Some experts are worried that giving AI companies free cloud credits could “lock in” those companies as customers, making it harder for them to switch providers or mix and match features from different ones. This would hurt competition.
Steven Weber, a professor at the Berkeley School of Information at the University of California, said at an FTC meeting this year, “The further down that road we go, the harder it is to loosen up.”
Some deals between cloud and AI companies, like the one between Amazon and Anthropic, can also give big tech companies the chance to buy shares in essential AI startups. In this way, the UK’s competition regulator is thinking about a possible antitrust investigation into Microsoft’s relationship with OpenAI. They say that OpenAI’s recent leadership problem may have led to a “relevant merger” that needs an investigation. Microsoft has said that it does not own “any portion” of OpenAI, but it did recently get a seat on the board of the startup without vote rights.
People are also questioning the fees that some cloud service providers charge for taking data out of their systems.
Prince explained that it only raises the cost of AI. There would be a big drop in the cost of AI training if companies could easily and cheaply move their data from one provider to another. “I think you could cut the cost of training probably in half,” one expert said.
How cloud providers defend their track record
Cloud providers and their supporters contend that the market is extremely competitive, despite the possibility that it is concentrated because of the prohibitive cost of developing cloud services on a massive scale.
According to Amazon, as part of the agency’s market assessment this summer, since AWS launched, its pricing for services like computer processing, data storage, and data transfer have considerably decreased. There is intense competition among cloud and other IT companies, which is good for the US economy.
Google last month unveiled a new generation of potent cloud-based computing processors that it claims can train huge AI models almost three times quicker than the previous generation, a move that is indicative of the company’s attempt to compete for clients using AI clouds.
Google claims that it trained Gemini, its most recent and advanced AI model, using such processors.
This year, Google stated in a blog post, “At Google Cloud, they are committed to being the most open hyperscale cloud provider, and that includes their AI ecosystem. Through collaborations, we can provide enterprises with easier access to and innovation with large language models and generative AI, apply rapidly developing AI and ML capabilities to real-world industry use cases, and create a new wave of applications that make use of all these skills.
Microsoft has stated that when they sign agreements with a cloud provider, cloud customers—including AI developers—are not naive.
According to Microsoft, customers negotiate a lot with service providers about terms including cost, storage capacity, length of contract, and other issues. This implies that every project is essentially a pitch for cloud vendors, as customers will look elsewhere if they do not receive the quality or value they anticipate from them. Cloud service providers must contend with this.
According to Brandon Jung, vice president of ecosystem and business development at AI company Tabnine and one of Google’s original cloud platform staff members, exclusive arrangements between AI startups and cloud providers can be beneficial.
The benefit is efficiency for developers and deployment, according to Jung, who also mentioned that using a single provider might improve security. However, he believes that it would further entwine people with the clouds.
The distrust that governments have long harbored toward app stores, social media, and e-commerce behemoths, according to critics, is the root cause of some of the issues lawmakers have brought up.
According to a person acquainted with the inner workings of large cloud suppliers who asked to remain anonymous due to not being permitted to talk in public, many of those are also the same organizations that manage enormous cloud computing businesses.
The individual stated that regulators worldwide are generally uneasy about Big Tech and digital markets, believing that the degree of concentration in these areas is unfavorable. Unfortunately, they’ve adopted some of the same worries and fears and incorporated them into their very different clients perception of cloud computing.
Certain AI firms have purposefully avoided taking on any exclusive relationships with cloud providers.
Although it has likely resulted in some financing opportunities being lost, Cohere AI, which offers AI models to business clients rather than consumers, stated that it collaborates with all major cloud providers.
Martin Kon, COO of Cohere, said, they were very deliberate about not taking any one check bigger than the others, during their most recent [financing] raise. They are not obligated to listen to anyone who claims to be the leader.
However, according to West, businesses such as Cohere are more the exception that proves the rule.
She stated that she believed it demonstrated how powerful cloud companies are in the industry.
