It isn’t your imagination; Google Cloud is flooding the zone

Spread the love

The100 billion The announcement between the Nvidia and the OpenAI, which was announced on Monday, represented the partnership-the latest mega-dill AI infrastructure reinstates the landscape. The agreement involves non-voting shares and adequate computing strengths involved in the purchase of a lot of chip for more than 5 million US families, deepening the relationship between two powerful players in AI.

Meanwhile, Google Cloud is a completely different bet. The largest players in the industry are hell about capturing the next generation of AI companies before the ever -strictly strict partnership cement.

Francis Desuza, its COO, has seen the AI ​​revolution from multiple vantage points. As a former Chief Executive Officer of the genomics giant Illuminar, he saw the invention of machine learning transform drug. Two-year-old AI alignment startup, as co-founder of Cynth Labs, has jumped on the protection challenges of increasingly strong models. Now, after joining the C-Suit in Google Cloud in January, he is doing a huge bet or orchestrate in the second wave of AI.

It likes to tell a story of Desija. In the conversation with this editor at the beginning of this weekHe mentioned several times that nine of the top 10 AI labs use Google’s infrastructure. He also said that almost all generators of AI unicorn in Google Cloud are driven, which all the worldwide general AI startups chose 5% of Google as their cloud suppliers, and the company has kept $ 58 billion in new revenue promises in the next two years, which represents more than its current annual run rate twice.

“The AI ​​Cloud Market is resetting the market, and Google Cloud is especially leading with startups,” he said in a soft-rich tone, masking an ambitious strategy to go home on delivery upstarts when the biggest players are busy with the largest players who are profitable but non-high.

The NVIDIA-Opnai Agreement gives an example of the scale of consolidation in the AI ​​infrastructure. Microsoft’s original $ 1 billion OpenAI investment has increased to about $ 14 billion, originally re -shape the cloud market. Amazon followed $ 8 billion for anthropological investment, securing deep hardware customizations that made AI training suitable for better work with Amazon infrastructure. Oracle has emerged as a wonderful winner, a landing a $ 30 billion Cloud Deal with Opena and then secured a jaw-dropping $ 300 billion The five -year promise begins in 2027.

Even the meta, despite its own infrastructure, has signed 10 billion dollars agreement The US infrastructure cost with Google Cloud has added another level to these inter -related partners involved in the US administration’s $ 500 billion “Stargate” project, softbank, openAI and Oracle when planning $ 600 billion out of 2028.

TechCrunch event

San Francisco
|
October 27-29, 2025

Companies like OpenAI and Nvidia may seem like a partnership that seems to be cementing elsewhere. In fact, it looks like Google is being deducted from some crazy dealmaking.

3D Google Logo
Google Logo Alfabet and Google CEO Beautiful Pichai and Polish Prime Minister Donald Task attend Google for Poland Warser Startups in Google on February 2025 for Poland Warser Startups (Figure Credit:Claudia Radeka / Nurfoto / Getty Figure

However, corporate Behemoth is not sitting in the hands of it. Instead, Google Cloud is signing smaller companies like loving and Windsorf – Desuja called “Next generation companies arrived” – without big investment as “primary computer partner”.

Reflects both the procedure opportunities and requirements. In a market where companies can “go to a startup in a very short time as a multi-billion dollar company”, as Deshuja said that future unicorns could prove more valuable than their giants before they matured.

The strategy extends beyond the general customer acquisition. Google provides AI Startups $ 350,000 on cloud credit, accessing its technical parties and helping the market through its market. Google Cloud also describes as a “any compromise” AI stack – from chips to model applications – with a “open policy” that gives customers the choice at each level.

“Companies like the fact that they can get access to our AI stack, they can get access to our team to understand where our technologies are going,” Desuja told us during interview. “They also like that they are getting access to the enterprise grade Google class infrastructure.”

This infrastructure has become more clear this month when the report revealed to expand the custom AI chip business behind the screen behind Google. According to informationGoogle has signed an agreement to keep the data center for the first time on its tensor processing unit (TPU), including $ 3.2 billion as a financial support for a New York facility, including an agreement with London -based Fluidstack.

At the same time, they need a fine to compete directly with AI companies when providing their infrastructure. Google Cloud Open provides TPU chips and hoses anthropic clode models through its Vertex AI platform, even competing from head to head with its own Gemini models. (Google Cloud’s main organization, Alfabet, also owns 14% of the ethnicity according to the New York Times Court documents, though when asked about Google’s financial relationship with anthropologists directly, Deshuja has called this relationship “Multi-Leward Partnership”.

However, if Google is trying to become Switzerland while moving its own agenda, it has a lot of practice. This method has Google’s open-source contribution roots, starting from Cubartus to Foundational “You need only” paper that enables the transformer architecture to the most modern AI. Recently, Google has published an open source protocol called Agent-to-Agent (A2A) for inter-agent communication in an attempt to show its continued commitment to the competitive field.

“We have created the obvious choice for years to expose at every level of the stack and we know that its meaning companies can absolutely take our technology and use it to create a competitor at the next level,” Desuja acknowledged. “It’s happening for decades that is what we’re fine.”

Google Cloud’s Startup’s Courtship comes in a particularly attractive moment. In this month, Federal Judge Amit Mehta gave a brief verdict in the government’s five -year -old search case, trying to prevent Google’s domination without interrupting its ambition.

Although Google avoided the most seriously proposed fines of the judiciary, including forced division of its Chrome browser, the company refers to regulatory concerns about the exclusive benefit of its investigation to dominate the ruling AI. Critics are concerned, understandable, the broad trove AI systems of Google search data provide an unjust benefit to developing and the agency may deploy the same exclusive strategies that secure its exploration dominance.

In the conversation, Desuja has focused on more positive results. For example, he says, “I think we have the opportunity to basically understand some major diseases that we do not only have a good idea today,” Desuja is outlined an outline where Google Cloud Alzheimers, Parkinson’s and Climate Technology help research. “We want to work very hard to make sure we are moving forward in the technology that we will be able to do” “

Critics cannot be easily assured. Locates itself as an open platform that empower the next generation of AI companies, instead of controlling the Google Cloud regulators to show that it encourages the competition instead of establishing a relationship with stipulates, which can help Google help Google.

For our full conversation with Desuja, see this week Downloading Strikelivies Podcast; A new episode is published every Tuesday.

Leave a Reply

Your email address will not be published. Required fields are marked *