Yazılar

OpenAI Partners with Google Cloud in Surprising AI Rivalry Deal

OpenAI has struck a significant cloud computing deal with Alphabet’s Google Cloud to support its growing AI infrastructure needs, sources told Reuters. This collaboration, finalized in May, marks an unprecedented partnership between two major competitors in artificial intelligence.

The move signals OpenAI’s efforts to diversify beyond its longtime partner Microsoft, which had exclusively provided data center services until January. Google Cloud will now supply additional computing power to OpenAI for training and running its large language models, including ChatGPT.

The deal highlights the immense compute demands required for AI development and how competitive dynamics are evolving. Despite the fierce rivalry—OpenAI’s ChatGPT poses a strong challenge to Google’s dominant search business—both companies have chosen to cooperate in meeting infrastructure needs.

Alphabet’s stock rose 2.1% following the news, while Microsoft shares slipped 0.6%. Analysts at Scotiabank called the partnership “somewhat surprising” but a strategic win for Google Cloud, which has been aggressively expanding its AI hardware offerings, including tensor processing units (TPUs) used internally and for other customers like Apple.

OpenAI’s recent moves to reduce dependency on Microsoft include partnerships with SoftBank, Oracle, and CoreWeave, as well as plans to develop its own AI chips to cut reliance on external hardware providers. Meanwhile, Microsoft and OpenAI continue to renegotiate their multibillion-dollar investment terms.

Google’s Cloud business, generating $43 billion in sales in 2024, aims to capture market share against rivals Amazon and Microsoft by positioning itself as a neutral cloud provider favored by AI startups with costly infrastructure needs.

This deal presents a complex balancing act for Alphabet CEO Sundar Pichai, who must allocate limited chip capacity between competing demands from Google’s own AI projects and cloud customers. Despite ChatGPT’s threat to Google’s search dominance, Pichai remains confident in the company’s position.

OpenAI’s Annualized Revenue Doubles to $10 Billion Amid AI Boom

OpenAI announced on Monday that its annualized revenue run rate surged to $10 billion as of June 2025, nearly doubling from about $5.5 billion in December 2024. This strong growth positions the company on track to meet its previously shared full-year revenue target of $12.7 billion.

The reported figure excludes licensing revenue from major backer Microsoft and large one-time deals, underscoring the core strength of OpenAI’s subscription and usage-based income from its AI models, including the widely popular ChatGPT.

Despite posting a loss of roughly $5 billion last year, OpenAI’s rapid revenue scale sets it well ahead of competitors. For comparison, Anthropic, another leading AI firm, recently surpassed $3 billion in annualized revenue fueled by demand from startups using its code-generation models.

OpenAI is also preparing for a major funding round of up to $40 billion led by SoftBank Group, valuing the company at $300 billion. Since launching ChatGPT over two years ago, OpenAI has expanded its offerings to include a variety of subscription plans for both consumers and businesses.

As of March 2025, OpenAI reported 500 million weekly active users, reflecting the broad and growing adoption of its artificial intelligence technology worldwide.

US-UAE AI Data Campus Deal Faces Delays Amid Security Concerns

A multi-billion dollar agreement to establish one of the world’s largest artificial intelligence data center hubs in the United Arab Emirates (UAE) remains far from finalized, according to sources familiar with the negotiations. Despite its high-profile announcement during President Donald Trump’s recent visit to Abu Dhabi, persistent U.S. security concerns continue to stall progress.

The planned 10-square-mile AI campus is being spearheaded by G42, an Emirati state-linked technology firm central to the UAE’s AI ambitions. Major U.S. technology firms including Nvidia, OpenAI, Cisco, Oracle, and Japan’s SoftBank have signed on to help develop the first phase, called Stargate UAE, which is scheduled to become operational in 2026.

The project’s backers have touted it as a significant step toward steering Gulf nations toward U.S. technology and away from Chinese alternatives. However, five sources involved in the discussions told Reuters that U.S. officials remain deeply concerned about potential technology transfers to China and the UAE’s ability to enforce strict export controls.

Although the UAE pledged during Trump’s visit to align its national security regulations with Washington — including measures to prevent diversion of U.S.-origin technology — American officials remain cautious. These concerns mirror those raised during both the Biden and Trump administrations, particularly over the UAE’s previous deployment of Huawei 5G infrastructure despite U.S. objections.

Sources indicated that the U.S. Commerce Department has yet to determine the security protocols required for exporting advanced Nvidia AI chips critical to the project. The absence of an agreed enforcement mechanism further complicates the deal, leaving it without a definitive timeline for completion.

Among the likely U.S. conditions are prohibitions on Chinese technology at the site and restrictions on employing Chinese nationals, given ongoing fears of AI chip smuggling and intellectual property leaks to adversaries. While the UAE has dismantled some Chinese partnerships—such as G42 removing Chinese hardware and divesting from certain Chinese holdings under Biden administration pressure—Chinese firms like Huawei and Alibaba Cloud still maintain a strong presence in the country.

Adding to U.S. unease is the UAE’s growing role as a hub for companies circumventing Western sanctions on Russia, further complicating Washington’s strategic calculus. Despite these challenges, both Trump administration officials and some in the current administration remain committed to pursuing the deal, though bipartisan skepticism remains strong in Congress.

Once operational, Stargate UAE is expected to house roughly 100,000 advanced Nvidia Grace Blackwell GB300 AI chips within a 1-gigawatt facility — potentially expanding to 5 gigawatts in the future. The Emirati government has so far not commented on the latest delays, and no final agreement has been reached on technology controls or operational oversight.