Google fights EU data-sharing as samsung makes $1bn AI infrastructure bet.
Regulators move to open data silos.
European regulators are directly targeting the data advantage of Big Tech in the AI era. In July, the EU introduced new Digital Markets Act (DMA) obligations requiring “gatekeeper” companies like Google to provide access to their data and platforms for rivals ([1]). This week Google escalated its resistance: it filed legal challenges to two EU orders that force it to share search data with competing AI developers and allow rival virtual assistants on its Android mobile platform ([2]). Google argues that these measures would undermine user privacy and security by forcing it to open sensitive search histories ([3]).
The EU sees access to data as essential for fair competition in a landscape where **search data** and usage signals have become a strategic asset for AI development. By pushing Google to open up, Europe aims to prevent dominant players from weaponising data as an AI monopoly. Under the DMA, regulators can impose fines up to **10%** of a company’s global turnover for non-compliance ([4]), reflecting how serious they are about breaking data silos. Google’s pushback highlights the tension between competition and privacy: it claims the mandated data-sharing lacks “sufficient anonymisation” and would cause “irreversible harm” ([5]). The coming court battle will test how far authorities can go in democratising access to high-value data without compromising user trust.
Big bets on AI infrastructure.
Meanwhile, industry leaders are pouring unprecedented investment into the physical side of AI. Samsung Electronics and five of its affiliates announced a **$1 billion** stake in Helix Digital Infrastructure - a new venture launched by KKR - to build AI-centric data centres that come bundled with their own power generation and fiber networks ([1]). The strategy: ensure that future AI models have dedicated, on-demand compute capacity and energy supply. Helix’s approach illustrates how AI infrastructure is evolving into a “massive, multi-sector ecosystem” spanning data centres, power grids, and communications networks ([2]), effectively treating data centres as core business platforms rather than peripheral IT costs.
This focus on custom AI infrastructure comes as the scale of AI operations skyrockets. In China, TikTok owner **ByteDance** has emerged as one of the largest consumers of computing power in the country, leasing capacity equivalent to roughly one-fifth of all China’s delivered data-centre power (from a national total of over **24 GW** and growing) ([3]). This makes ByteDance the single biggest data-centre tenant in China’s cloud market. Such numbers underscore that AI supremacy now requires enormous hardware and energy commitments: data-centre capacity "has become one of the clearest proxies for strategic AI ambition" ([4]). In other words, AI’s competitive frontier is increasingly an **infrastructure arms race**, where scaling model training and deployment hinges on controlling computing resources at unprecedented levels.
Data governance and sovereignty take centre stage.
As enterprises push AI deeper into their operations, they are discovering that data governance can be a make-or-break factor. Many organisations have rapidly deployed generative AI tools and "agent" applications that act autonomously, but few have full visibility into how these AI systems access and use internal data. In one cautionary example, a security review at a large U.S. company found **85,000** internal files that were inadvertently left exposed to AI-powered tools and agents ([1]). Such unchecked "AI sprawl" raises obvious security and compliance risks.
The result is a surge in demand for new solutions to monitor and control data used by AI. This week saw major funding for startups in the emerging “AI agent” governance space. Reco, a U.S.-based AI security firm, raised **$55 million** in new funding from investors including AT&T ([2]), aiming to help companies track and manage their proliferating AI bots. Reco’s platform integrates with hundreds of apps to detect what software agents are doing and restrict their access to sensitive databases and files, offering a much-needed layer of oversight as AI becomes entangled in everyday workflows.
Beyond internal controls, companies are also rethinking where their AI lives to meet regulatory and ethical demands. **Data sovereignty** is becoming a top priority, especially in regions with strict privacy laws. This week in India, IBM partnered with local data services firm Yotta to launch a “sovereign” AI cloud platform for Indian organisations ([3]). The service, built on IBM’s watsonx technology and Yotta’s cloud, allows enterprises to deploy AI models and agents entirely within India’s own data centres, ensuring all data and operations remain under domestic governance. As regulations such as Europe’s forthcoming AI Act put pressure on data control, businesses are pre-emptively investing in robust data governance frameworks and localised infrastructure. The takeaway for CDOs and CTOs is clear: whether it’s containing AI internally or complying with national data laws, strong data architecture and governance practices are now essential to scale AI with confidence.