Oracle's Rise: The New AI Infrastructure Order

Published15 Sept 2025

Coverage8 Sept 2025 – 15 Sept 2025

Stories tracked96

Featured5

AuthorPulse24 Desk

Last updated15 Sept 2025

This week’s pulse

The abstract structures of the AI industry are solidifying into hard legal agreements, new infrastructure hierarchies, and concrete regulations. This week is about the formalisation of power, as OpenAI restructures its corporate entity, Oracle ascends as a kingmaker, and publishers demand payment. For decision-makers, the challenge is navigating a landscape where the rules are finally being written, not just imagined.

The Great Restructuring: OpenAI and Microsoft Redefine Their Pact

What happened

OpenAI is nearing a deal with Microsoft to restructure into a for-profit benefit corporation. The move is designed to pave a path towards a future IPO. Microsoft is renegotiating its partnership to ensure continued access to OpenAI's technology, even after AGI is hypothetically achieved. The new terms would see Microsoft gain long-term tech access in exchange for adjusting its equity stake in a restructured OpenAI.

So what

This is the financial and legal formalisation of the AI race's endgame. It signals a critical shift from ambiguous partnerships to concrete governance, de-risking technology access for hyperscalers while creating a viable capital structure for labs to fund the immense cost of AGI development. The operating model for foundational AI is being rewritten.

Oracle's Coup: A New Kingmaker in AI Infrastructure

What happened

Oracle's market value and future bookings have skyrocketed. The surge is driven by massive new cloud infrastructure deals. This includes a landmark agreement with OpenAI, reportedly worth up to $300 billion over five years. The demand has propelled Oracle's remaining performance obligations to $455 billion, a 359% year-over-year increase.

So what

This catapults Oracle into the top tier of AI infrastructure providers, directly challenging the established dominance of AWS, Azure, and GCP. It proves that specialised, large-scale compute capacity is a commodity so valuable it can reorder the cloud market, creating a new centre of gravity for AI development.

03Pay or Perish

Pay or Perish: The Copyright War Enters the Licensing Era

What happened

The battle over training data is escalating and formalising. Penske Media is suing Google for using its content in AI Overviews. This follows Anthropic agreeing to a $1.5 billion settlement in a copyright lawsuit with authors. In response, a new framework called Real Simple Licensing (RSL) is emerging to streamline data deals between publishers and AI firms.

So what

The market is bifurcating between costly litigation and the creation of new licensing mechanisms. This forces a strategic choice for AI builders: budget for data as a direct operational cost or risk expensive, brand-damaging legal battles. The era of treating public web data as a free resource is definitively over.

Microsoft's Hedge: The End of Single-Provider AI Stacks

What happened

Microsoft is strategically diversifying its AI suppliers. The company is integrating models from Anthropic into its Office 365 suite. This move reduces its deep reliance on its primary partner, OpenAI. The decision was reportedly influenced by internal tests where Anthropic's models outperformed in certain tasks.

So what

This is a crucial de-risking strategy from the world's largest software company. It signals that even the biggest players will not bet on a single model provider. For enterprise architects, this validates a multi-cloud, multi-model approach to avoid vendor lock-in and leverage best-in-class performance for specific use cases.

California Sets the Standard: AI Regulation Becomes Real

What happened

California's legislature has passed a landmark AI safety and transparency bill (SB 53). The bill mandates that large AI firms publish safety frameworks and report critical incidents. This comes as the FTC launches a formal inquiry into the risks AI chatbots pose to children, ordering seven top developers to provide data on their safety measures.

So what

While federal action stalls, California is creating a de facto national standard for AI governance. This establishes a complex compliance landscape for builders, who must now navigate a patchwork of state-level rules that will likely influence future federal legislation and define the boundaries of acceptable risk.

⚡ Quick picks

Faster moves.

Markets 💹: TSMC reported a 34% revenue increase for August, demonstrating sustained global demand for the advanced chips powering the AI boom.

Finance 💷: Quantum computing firm PsiQuantum secured $1 billion in Series E funding, with participation from Nvidia's venture arm, to build its fault-tolerant photonic quantum computer.

Risk ⚠️: A North Korean hacking group successfully used ChatGPT to create a deepfake military identification document in an attempted cyberattack, highlighting the weaponisation of publicly available AI tools.

Macro 🌍: Nvidia is facing increased regulatory pressure in China after an initial probe found the company in violation of anti-monopoly laws, adding complexity to US-China tech relations.

Pulse24’s view

The AI industry's 'move fast and break things' era is ending. It is being replaced by a period of intense structural negotiation over governance, infrastructure control, and data rights. The wild frontier is being carved into territories with defined borders and tolls. For leaders, the challenge is shifting from pure innovation to strategic positioning within this newly ordered world. Are your partnerships and compliance frameworks built for this new reality?