Amazon is maintaining advanced talks to inject around $10.000 billion into OpenAI, the company responsible for ChatGPT, in a deal that could put its valuation above $500.000 billion and consolidate it as one of the most valuable technology startups on the planet.
The negotiations, which various media outlets such as The InformationCNBC and Reuters describe it as fluid and not yet closedThey would not be limited to a simple financial transaction: the plan also involves strengthening technological collaboration between the two companies, reinforcing the using the Amazon Web Services (AWS) cloud and the adoption of Trainium artificial intelligence chips by OpenAI.
A multi-million dollar investment and a record valuation
Sources familiar with the process indicate that Amazon is studying commit more than 10.000 billion dollars at OpenAI, a figure that some leaks place even above that threshold, which would fit within a broader funding round with other institutional investors.
This capital injection could to raise OpenAI's valuation to, or even above, $500.000 billion, reinforcing the perception of the company as a central driver of the current wave of generative artificial intelligence, despite the doubts of some analysts about a possible bubble in the sector.
Investor interest is based on the fact that OpenAI has closed in recent months major secondary stock salesallowing current and former employees to divest at valuations already close to that figure, which has occasionally placed it above SpaceX as the world's most highly valued startup.
Within this context, various investment funds and Wall Street banks They have warned about the “circular” nature of some agreements, in which large technology companies invest billions in potential customers with the aim of ensuring that those same customers consume their cloud infrastructure or their AI chips on a large scale.
A strategic shift following the restructuring of OpenAI
The talks between Amazon and OpenAI come at a key moment for the company led by Sam Altman, which It completed a profound restructuring of its governance last October, effectively abandoning the non-profit foundation model as the sole focus and giving more weight to the for-profit structure.
That change has OpenAI granted greater flexibility to raise capital and forging alliances with various major technology companies, which has resulted in a network of agreements that goes far beyond its long-standing main partner, Microsoft.
Microsoft has invested more than 2019 $13.000 billion in OpenAI and, according to various sources, owns around 27% of the company's economic capital, with preferential marketing rights for certain advanced models through Azure.
However, the new framework agreed last fall It eliminated Microsoft's right of first refusal to be the exclusive provider of computing.opening the door for OpenAI to deploy some of its products and workloads on other cloud providers such as Amazon or Oracle, always respecting certain exclusivity commitments on API products linked to Azure.
This relaxation of exclusivity has been key for OpenAI to be able to to distribute their enormous computing needs among several strategic partners, reduce their dependence on a single cloud and improve their bargaining power on prices and access to chips.
AWS enters the scene: a $38.000 billion mega-contract
Even before discussing capital, OpenAI and Amazon had already sealed a huge agreement in the field of infrastructure: a contract worth approximately $38.000 billion to acquire computing capacity on AWS for a period of approximately seven years.
This agreement makes Amazon Web Services one of the major computing providers for OpenAIThis move marks the end of almost exclusive reliance on Microsoft's cloud and places AWS at the center of the race to power the most advanced models.
In the first phase, OpenAI will run its workloads in existing AWS data centers, taking advantage hundreds of thousands of Nvidia GPUs deployed in US territory. Over time, Amazon plans to expand Additional infrastructure specifically for OpenAIThis involves building and adapting custom data centers to meet the company's growing computing demands.
This mega-contract with AWS adds to other agreements OpenAI has with Providers such as Oracle and Google Cloud, as part of a deliberate effort to diversify cloud providers and mitigate risks related to capacity, chip supply, and pricing in an extremely strained market.
For Amazon, the alliance means reaffirm AWS's role as the dominant cloud operator compared to Azure and Google Cloud, at a time when their cloud revenue growth rates had fallen somewhat behind those of their major competitors.
Trainium and Amazon's own chips: the key technical component
Beyond cloud contracts, much of Amazon's interest lies in getting OpenAI to adopt Trainium artificial intelligence chipsdesigned to train and serve large-scale AI models on AWS infrastructure.
Amazon carries developing its own line of AI processors since around 2015AWS has developed the Inferentia chips, primarily focused on inference, and more recently, the Trainium family for intensive model training. The latest generation of Trainium was unveiled by AWS earlier this month.
The company's message is clear: Trainium aims to offer a better price-performance ratio than traditional Nvidia GPUs, providing customers with a more efficient alternative for very cost- and energy-intensive workloads.
AWS executives like Dave Brown have insisted that these chips They seek to broaden the range of options for customersso that they can combine Nvidia GPUs with Amazon-specific hardware based on their specific needs for cost, latency, and scalability.
If OpenAI were to significantly increase its use of Trainium for any part of its models, the market would interpret the move as a boost to Amazon's semiconductor strategywhich until now had had more difficulty gaining massive traction against Nvidia's unstoppable dominance in the field of generative AI.
Infrastructure commitments and the race for chips
Amazon's potential investment comes in a context of OpenAI's enormous infrastructure spending commitments, which would have closed deals worth more than $1,4 trillion related to chips, data centers and energy to power its models.
These commitments include contracts with semiconductor manufacturers such as Nvidia, AMD, and Broadcomas well as alliances with major cloud providers and large-scale data center projects, including developments such as the well-known Stargate project with Oracle, aimed at deploying multi-gigawatt infrastructures.
In this scenario, opting for alternative chips like Trainium doesn't mean abandoning Nvidia GPUs, but configure a “multichip” ecosystem in which different architectures coexist to respond to a computing demand that is growing at an almost exponential rate.
Analysts emphasize that ensure a stable supply of chips and power It has become one of the most critical factors for the survival and leadership of large AI laboratories, often above even pure algorithmic innovation.
Meanwhile, major technology companies such as Meta and Google They are also looking for alternatives or complements to Nvidia, either with their own chip designs or by resorting to third-party offers, in an attempt to reduce dependence and gain room for maneuver in costs.
The role of Anthropic and the war for generative AI
Amazon's potential investment in OpenAI has an additional interpretation: Jeff Bezos' company has already invested at least $8.000 billion in Anthropic, one of OpenAI's main rivals in the development of large language models, responsible for the Claude family.
That participation has turned AWS into preferred infrastructure provider for Anthropicwhich also uses Trainium and Inferentia chips in some of its workloads, fitting into Amazon's strategy of promoting its hardware as an alternative to leading GPUs.
Microsoft has followed a similar path with Anthropic, recently announcing that It could invest up to an additional $5.000 billion in the startup, while Nvidia has reportedly committed up to $10.000 billion to the same company, in a move that illustrates the intensity with which major players are positioning themselves around the most advanced AI labs.
In this context, Amazon's potential involvement in OpenAI is interpreted as a hedging strategy: being simultaneously exposed to several of the potential winners in the generative AI race, while boosting the use of its own cloud infrastructure and chips.
For OpenAI, adding Amazon to its list of major partners, alongside Microsoft, Nvidia, and Oracle, means strengthen its financial muscle and ensure that it can continue to finance the development of increasingly expensive and complex models.
Impact on the market and doubts about a possible bubble
The announcements of agreements between Amazon and OpenAI have already had an effect on the markets: Amazon's shares have reached all-time highs after the announcement of the $38.000 billion mega-contract with AWS and the rumors about double-digit billion-dollar investments.
Investor enthusiasm is explained by the expectation that AWS captures an even larger share of AI infrastructure growth, at a time when the cloud business is one of the main drivers of profits for the group.
However, there are voices calling for caution: some Wall Street analysts have warned that valuations of major AI companies and the volume of committed capital They might not correspond to the tangible benefits that the sector currently generates.
In particular, the risk is noted that cross-investments between technology companies and AI labs —where a cloud provider finances a client who, in turn, commits to buying billions in computing capacity— could lead to a bubble that is difficult to sustain if revenues from commercial products do not grow at the expected rate.
Despite these reservations, the big tech companies continue to make moves and announcing multi-year spending and investment plans, convinced that generative AI and foundational models will be a key pillar of their businesses in the next decade.
Europe and Spain face the new wave of alliances in AI
Although the negotiations between Amazon and OpenAI are mainly taking place in the United States, Its effects will also be felt in Europe and, by extension, in Spain.both in terms of access to technology and in terms of regulation and competition.
On the one hand, the consolidation of alliances between large cloud providers and AI labs points to a greater concentration of power in a few multinational corporationsThis is something that worries European authorities within the framework of regulations such as the recently approved Artificial Intelligence Act (AI Act) and the regulation of digital markets.
In parallel, European companies and public administrations They rely heavily on cloud services from players like AWS, Azure, or Google Cloud to deploy AI solutions, which may intensify the debate on digital sovereignty and the need for own infrastructures in the region.
For Spanish companies already working with AWS—from large banks and energy companies to technology startups—OpenAI Expand your deployment on the Amazon cloud This can translate into more direct and efficient access to advanced models through simplified integrations within the cloud platform itself.
At the same time, the heavy concentration of computing resources in a few global providers raises questions about Europe's ability to compete in the development of their own foundational models, beyond specific alliances or projects driven by public and private consortia.
In any case, the eventual closing of the agreement between Amazon and OpenAI It would redefine the power map in artificial intelligenceaccelerating the convergence between large clouds, chip manufacturers and modeling labs, and forcing European regulators, companies and users to adapt to an increasingly concentrated and resource-intensive environment.
With everything already on the table—mega-computing contracts, investments of tens of billions, bets on proprietary chips, and cross-deals between tech giants—the negotiation for Amazon to acquire a stake in OpenAI is shaping up to be one of the most relevant moves in this new phase of the AI racewhich can determine for years who controls the infrastructure, the technology and, ultimately, the business associated with generative artificial intelligence models.
