In the realm where AI supremacy is up for grabs, here are some key battlegrounds:
First, there's the talent required to design chatbot models for artificial intelligence. Moreover, the AI model must process numerous training data to learn how it should react to your prompts. Thirdly, advanced silicon chips come into play for training sessions, a process that can take months even in the fastest chips.
Cloud platforms are rapidly becoming the fourth essential component in the AI spectrum. They collect data from various chips, offer online storage, and provide numerous other services. They also lease computing capacity to AI companies in need of raw computational power, while storing training data wherever it is needed.
The dominance of AI developers on cloud services dictates the rise and fall of the entire AI sector, placing cloud providers at the center of a technology that promises to reshape the way humans work, play, and learn.
Currently, a handful of heavyweights dominate the $5 trillion cloud market, including Amazon, Microsoft, and Google. Concerns are now being raised about Big Tech's excessive influence on the cloud market, as it could potentially have far-reaching and potentially anticompetitive impacts on the future of AI.
Senator Elizabeth Warren, D-Mass., stated to CNN: "I am deeply concerned that a few large technology companies dominate Cloud Computing and Storage." She warned that, "Without reasonable regulation, these companies will only strengthen their dominance in the AI sector, destroying competition, jeopardizing consumer privacy and security, stifling innovation, and threatening national security."
All three companies refused to comment on this article. Technology industry defenders argue that the leading companies in the sector are fierce competitors, and their size helps them provide comprehensive AI solutions to meet their clients' various Cloud Computing requirements.
However, political decision-makers, like in any large industry controlled by mega-corporations, focus on the potential for monopolistic pricing, anticompetitive agreements, exploitative contract terms, or other practices that could influence whether AI services are used or the eventual cost of AI products on the market.
The cloud market is limited to a few major players
As generative AI gathers steam, expectations are high for its share of overall Cloud expenses to significantly increase.
While the Public Cloud market typically grows exponentially, with estimates from market research firm Gartner suggesting the overall expenses will climb more than 20% this year to hit $679 billion, AI is expected to constitute 30% to 50% of that increase.
Matthew Prince, CEO of Cloudflare, asserted that only a handful of Cloud platforms can provide the massive processing power needed by companies and individual users as AI use proliferates.
The Cloud Market is currently only a small fraction of the $563 billion Public Cloud market, with AI composing less than 10%. Amazon, Microsoft, and Google lead the way, with smaller Cloud providers like IBM and Oracle following close behind.
Regulatory scrutiny is intensifying worldwide
Governments worldwide are paying closer attention to developments in this increasingly critical sector.
In the United States, both the Federal Trade Commission (FTC) and President Joe Biden have expressed concerns about the competitive dynamic in the cloud market, which could stifle the growth of AI.
In June, the FTC released a statement that, should one or more companies gain control of the Cloud market or make significant contributions to AI development, they "could have an overbearing influence on substantial economic activities." As AI becomes increasingly embedded in everyday life, it will potentially grow into a fundamental aspect of people's daily lives.
Biden reiterated these concerns in October by signing an executive order against AI, which states that fostering fair and open markets for AI development requires combating "illegal business practices" and addressing the risks posed by the use of critical assets like semiconductors, computing power, storage, and data by dominant companies that could disadvantage competitors.
Though the order does not specifically name any companies and does not implicate the industry with any wrongdoing, the industry has been under the microscope from various regulatory bodies in numerous countries in recent years.
For instance, a UK government investigation focused on Microsoft's software licensing practices (with little relation to AI), while the FTC announced it was examining the sector alongside France, Japan, the Netherlands, and South Korea in March. Many of these investigations have concluded that the market is highly concentrated, with Sarah Myers West, director of the AI Now Institute and former FTC advisor for AI, claiming that "the vast majority of the investigations carry the same finding – that this is a highly concentrated market."
The reason cloud platforms partner with AI companies is that when their products eventually reach consumers or businesses, there is still a need for more Cloud resources. Such collaborations could also allow major tech companies to collect more data from businesses and consumers.
"They want to introduce people into their Cloud ecosystem, which is the most direct way to monetize these tools," West stated.
What worries Cloud critics
Some industry analysts caution against specific Cloud practices that could raise anticompetitive concerns.
For example, many AI companies are required to sign exclusive agreements with a single or two Cloud providers as a condition for significant investments. Exclusive offers in the Cloud market have become standard practice for AI developers, in contrast to other parts of the industry where multi-provider usage arguably occurs more frequently.
West, the AI Now Institute director, argued that "many of the currently active AI startups developing tools have already signed exclusive licensing agreements with Cloud providers." Even smaller providers often work through this channel and usually only through one Cloud provider.
As an example, consider OpenAI. Its ChatGPT tool, which gained popularity in 2021 and ignited the current AI trend, relies heavily on a partnership with Microsoft, which offered it $13 billion in cash and free Cloud Computing services.
Satya Nadella, Microsoft CEO, discussed his company's role in November with CNN Tech reporter Kara Swisher: "Without the extensive support of Microsoft and the partnership with this company to achieve its mission, there would be no OpenAI." He added that this would not be possible under the current regulatory environment. During the OpenAI's crisis, Microsoft was able to take complete control of the startup's entire AI work. "We have the talent, we have the computing power, we have the data, we have everything."
Early in the year, Amazon announced that it would invest up to $4 billion in AI startup Anthropic in exchange for a minority stake and the title of a "major" Cloud provider. Previously, Anthropic announced a deal to make Google its "preferred" Cloud provider.
Some experts warn that cost-free Cloud credits could enable powerful Cloud providers to "bind" AI companies as customers and stifle competition by making it difficult for them to switch providers or combine services from competing providers.
"The more we move in this direction, the more difficult it will be to unwind," said Steven Weber, a professor at the University of California, Berkeley School of Information, in February's FTC Roundtable.
Deals between Cloud Computing companies and AI firms like Amazon and Anthropic could also provide the former with stakes in influential AI startups. The British Competition and Markets Authority is considering launching a cartel investigation into Microsoft's relationship with OpenAI and has hinted that the recent management crisis at OpenAI could have led to a "relevant merge," which merits examination. Microsoft denies owning "any part" of OpenAI, although it recently acquired a non-voting seat on the startup's board.
Another concern is the fees charged by some providers for data extraction from Cloud services.
"It makes AI more expensive," Prince said. If AI companies could easily and cost-effectively move their data between providers, "training costs for AI would be significantly reduced … I estimate they could cut training costs in half."
Cloud providers and their supporters argue that the Cloud market is fiercely competitive, despite the high costs associated with building comprehensive Cloud services in large scale.
"Since AWS was introduced, prices for services like computer processing, data storage, and data transfer have drastically dropped," Amazon told the FTC in the summer's Branchenstudie of the agency. "Competition between Cloud Computing and other IT providers is increasing, which is good for the US economy."
Google launched its latest generation of cloud-based computing processors this month, which it claims can train large AI models almost three times faster than previous generations. This demonstrates its commitment to competing in the cloud market for AI.
Google stated: "We at Google Cloud are committed to being the most open Hyperscale Cloud provider, and this includes our AI ecosystem." By partnering with companies, we can help them "easily access generative AI and large language models, quickly develop AI and machine learning applications to solve real-world business challenges, and leverage emerging AI and machine learning capabilities to build a new wave of applications."
Microsoft asserted that its Cloud customers, including AI developers, are not naive when it comes to negotiating contracts with Cloud providers.
"Customers negotiate a wide range of topics with providers, including price, storage capacity, contract term length, and more," Microsoft told the FTC. "This means that, for the Cloud provider, every project is also an opportunity to pitch – if a customer doesn't receive the required quality or value from the Cloud provider, they will switch to other alternatives."
Brandon Jung, VP for Ecosystem and Business Development at AI startup Tabnine and one of the earliest Cloud Plattform employees at Google, said that exclusive agreements between AI companies and Cloud providers could be beneficial.
"The advantage lies in developer and deployment efficiency," Jung stated, adding that using a single provider also increases security. "However, I think it will bind users more tightly to Cloud."
Some critics argue that some of the concerns raised by political decision-makers are a byproduct of the long-standing mistrust of big tech, social media platforms, and e-commerce giants. A source familiar with the thinking of major cloud providers said they are essentially tech companies that happen to offer Cloud services, rather than the other way around.
"Regulatory bodies worldwide are generally concerned about large tech companies and digital markets and see concentration in these markets as a problem," the source commented. "Unfortunately, they share some of the same concerns in the Cloud Computing and Cloud Client segment – Cloud Computing and Cloud clients are very different from each other."
However, some AI companies deliberately avoid exclusive arrangements with Cloud providers.
Cohere AI, which offers AI models for enterprise clients rather than consumers, says it works with all major Cloud providers despite the risk of missing out on some financing opportunities.
"During our last funding round, we specifically aimed to raise a lower, non-differentiating sum compared to other rounds," Martin Kon, Cohere's COO, stated. "We would not appreciate if someone were to be seen as the 'leader'."
However, Jung, from Tabnine, stated that exclusive agreements could be beneficial.
"The advantage is in developer and deployment efficiency," Jung added. "Using a single provider also increases security. However, I think it will bind users more tightly to Cloud."
There is a growing understanding of the need to regulate AI, with many nations already discussing and implementing policies to ensure ethical use of the technology. In the Netherlands, for example, a recent study by the Amsterdam AI Coalition found that 57% of surveyed market participants believe that the government should take a more active role in regulating AI.
In conclusion, the battle for AI supremacy hinges on several factors, including technological talent, advanced silicon chips, and the cloud platforms that enable AI development and deployment. As AI continues to reshape industries and economies, the competition for dominance in these areas will only intensify.
Source:
Enrichment Data:
The dominance of cloud providers is a critical factor in the competition for AI development. Cloud providers offer the necessary infrastructure, scalability, and accessibility for companies to innovate and deploy AI solutions. The major cloud providers--Amazon Web Services (AWS), Microsoft, and Google--invest heavily in AI infrastructure, leading the way in the development of cutting-edge AI technologies. This includes building new data centers, enhancing existing infrastructure, and investing in AI research and development.
The global AI landscape is becoming increasingly competitive, with players like Chinese firm DeepMind driving advancements fueled by generous government support. The rise of AI has led to a surge in cloud spending, which is expected to continue as businesses and individuals embrace AI-powered solutions. In response, cloud providers are investing billions of dollars in AI research, development, and infrastructure to maintain their competitive edge and support their customers' AI initiatives.
Regulators worldwide are paying closer attention to the AI sector, with concerns about data privacy, ethical use of AI, and anticompetitive practices. Governments are under pressure to ensure that AI is developed and deployed responsibly, and many are currently discussing or implementing policies to address these concerns.
The future of AI competition remains unclear, as new technologies and advances continue to emerge. However, cloud providers are expected to play a significant role in this competition, given their dominant position in the sector and their commitment to AI research, development, and deployment.
Sources: [1] Amazon Web Services, [2] Microsoft Azure, [3] Google Cloud Platform, [4] DeepMind, [5] Intel, [6] NVIDIA.
