The $ 100 billion partnership between Nvidia and Openai, announced Monday, represents-for now-the latest mega-deal remodeles the AI infrastructure landscape. The deal includes shares that are not linked to votes associated with mass chips and enough computer science for more than 5 million US households, deepening the relationship between two of AI’s strongest players.
Meanwhile, Google Cloud places a different bet completely. While the biggest players in the cement industry that are always lighter partnerships, Google is Hellbent to record the next generation of AI companies before they become too long in court.
Francis Desouza, his Coo, has seen the AI revolution from multiple advantages. As a former CEO of Genomics Giant Illumina, he watched the discovery of drug transformation transformation. As a co -founder of a two -year -old AI alignment start, SYNTH LABS has faced the security challenges of the ever -powerful models. Now, having joined the C-Suite at Google Cloud in January, she orchestrates a huge bet on the second wave of AI.
It’s a story that Desouza likes to say in numbers. In a conversation with this editor earlier this weekHe noted several times that nine of the top 10 AI laboratories use Google infrastructure. He also said that almost all AI Unicorns’s genetic unicorns are running on Google Cloud, that 60% of all newly established Gen AI companies worldwide chose Google as a cloud provider and that the company has ordered $ 58 billion in new revenue commitments over the next two years, which is more than two years.
Asked what percentage of Google Cloud revenue comes from AI companies, instead offered that “AI reinstates the Cloud market and Google Cloud drives the road, especially with newly established businesses”.
The NVIDIA-EPENAI agreement is an example of the scale of the AI Infrastructure integration. Microsoft’s initial Openai investment has increased to nearly $ 14 billion, fundamentally reforming the Cloud market. Amazon followed $ 8 billion in anthropogenic investments, ensuring deep material adjustments that essentially adapt AI training to work better with Amazon’s infrastructure. Oracle has emerged as a surprise winner, landed a $ 30 billion cloud agreement with Openai and then securing a five -year $ 300 billion commitment launched in 2027.
Even Meta, despite building its own infrastructure, signed a $ 10 billion deal with Google Cloud, planning $ 600 billion in US infrastructure spending by 2028.
TechCrunch event
Francisco
|
27-29 October 2025
These giant agreements may seem threatening to Google, given the corporate relationship that companies such as Openai and Nvidia seem to sting elsewhere. In fact, it looks very much like Google cuts from some frenzied negotiation.
But the corporate magnificent is not exactly sitting in his hands. Instead, Google Cloud signs smaller companies such as Lovable and Windsurf – what Desouza calls “the next generation of companies coming” – as “prime computers” without significant investments.
The approach reflects both the opportunity and the necessity. In a market where companies can “start being a company of many billions of dollars in a very short time”, as Desouza puts it, recording the future unicorn before maturing they prove to be more valuable than fighting today’s giants.
The strategy extends beyond the simple acquisition of customers. Google offers the newly established AI 350,000 companies in cloud credits, access to its technical teams and support for market market through its market. Google Cloud also provides what Desouza describes as “no compromise” stack AI – from brands to models to applications – with an “open ethos” that gives customers to choose each layer.
“Companies love the fact that they can access our stack AI. They can access our teams to understand where our technologies are going,” Desouza said during our interview. “They also love that they have access to the Google Class Infrastructure business.”
This infrastructure advantage became more apparent this month, when the report revealed that Google behind the stage maneuvering to expand the custom business AI Chip. According to informationGoogle has reached agreements to install the Tennter Editor (TPU) in other cloud providers’ centers for the first time, including a London -based Fluidstack agreement, which includes up to $ 3.2 billion in financial support for a new York facility.
Competition directly with AI companies while providing their infrastructure requires refinement. The Google Cloud provides TPU chips in Openai and hosts Anthropic’s Claude model via the Vertex AI platform, even when its own Gemini models compete with both. (Google Cloud’s parent company, Alphabet, also holds a 14% share of mankind, according to New York Times Court documents acquired earlier this year, though when he was immediately asked about Google’s financial relationship with the man, Desouza calls on the relationship that can access various fundamental models.
But if Google is trying to be Switzerland, while promoting its own agenda, it had many practices. Approach is rooted in Google’s open source contributions, from Kubernetes to fundamental “Attention is all you need“The document that allowed the architecture of the transformer, which is the basis of the most modern AI. More recently, Google has published an open source protocol called Agent-To-AGent (A2A) for communication between the agent in an effort to prove its continuing commitment to openness.
“We have made the explicit choice with the years open to every layer of the stack and we know that this means that companies can take our technology perfectly and use it to build a competitor in the next layer,” Desouza acknowledged. “This has been happening for decades. This is something we are okay.”
The Google Cloud Courtship for the newly established businesses comes at a very interesting moment. As soon as this month, federal judge Amit Mehta has made a separate decision to the five -year search for the government’s search, trying to limit Google’s sovereignty without preventing AI’s ambitions.
While Google avoided the most serious proposed sanctions of the Ministry of Justice, including the forced divestment of the Chrome browser, the decision underlined regulatory concerns about the company by utilizing the search monopoly to dominate AI. Critics are concerned, understandable, that the huge hometown of search data provides an unfair advantage in developing AI systems and that the company could develop the same monopoly tactics that have secured the domination of the search.
In the conversation, Desouza focuses on much more positive results. “I think we have the opportunity to fundamentally understand some of the most important diseases that today do not have a good understanding,” Desouza said, for example, describing a vision where Google Cloud helps research in power in Alzheimer’s technologies, Parkinson. “We want to work very hard to make sure we pioneer in the technologies that will allow this project.”
Critics may not be easily overturned. By placing it as an open platform that authorizes and not controls the next generation of AI companies, the Google Cloud can show regulatory principles that encourage competition rather than stifling it, while hammering relationships with newly established businesses that could help google if the google is increasing.
For our full chat with Desouza, check the week this week Strictlyvc download Podcast? A new episode comes out every Tuesday.
