The Battlefield In The War For Data Control

AI’s Future Depends on Who Controls the Data

Meta’s chief AI scientist, Yann LeCun, speaking at Davos, predicted that today’s generative AI models—large language models (LLMs)—will soon be obsolete. He argues that we are on the verge of a new AI paradigm that will move beyond simple pattern recognition into actual reasoning, planning, and real-world understanding. He also sees the coming decade as the “decade of robotics,” where AI systems will not just process information but interact with the physical world in unprecedented ways.

LeCun is correct that AI is evolving fast, and we’re about to witness a transformation that makes today’s capabilities look primitive. However, he underplays a critical factor: the most significant barrier to this future isn’t just computing power—it’s data control. The real battle in AI won’t be over GPUs alone; it will be over who controls the training data that fuels intelligence in the first place.

This is a growing concern among policymakers and industry leaders alike. Benoît Cœuré, president of France’s Autorité de la Concurrence (Competition Authority), recently warned that AI dominance relies on the control of data and computing power, concentrated in a handful of U.S. tech giants. He put it bluntly to Le Monde:

AI technology relies heavily on data, computing power, and talent, which are areas dominated by a few powerful American companies.

AI monopolies are the new oil cartels if data is the new oil. As AI shifts toward reasoning-based models, the question isn’t just how powerful AI will become but who owns and controls the data that fuels it.

The False Promise of De-Identified Data in AI

For years, businesses have relied on the idea that de-identified data safeguards against privacy concerns. The logic goes that if you strip away personally identifiable information (PII), the data can be safely used for analytics, advertising, and AI training without violating privacy.

But this argument collapses when faced with reality. The LoanDepot breach in 2024 is a perfect example. The mortgage lender suffered a massive data leak affecting 16.9 million customers, exposing names, addresses, financial account numbers, phone numbers, and dates of birth. While some argue that once “de-identified,” data is harmless, breaches like this prove otherwise—re-identification is often trivial. Financial data is a goldmine for fraudsters and AI models alike.

Dr. Courtney C. Radsch, Director of the Center for Journalism and Liberty and a nonresident senior fellow at Brookings, the Center for International Governance Innovation, and the Center for Democracy and Technology, in an article for Tech Policy Press, recently emphasized the need to dismantle data monopolies before it’s too late:

Unless we dismantle these data monopolies and encourage practices that protect privacy and competition, innovation will be little more than a hallucination that benefits dominant incumbents at the expense of citizen and consumer welfare, choice, and rights.

Visa, the only company effectively managing de-identified data at scale, is now being sued for monopoly practices. This raises the question: if even the most sophisticated financial data management players face legal scrutiny, what happens when AI-driven entities take over?

AI’s Next Leap Will Multiply Data Security Risks

We’re on the brink of three fundamental shifts in AI that will supercharge innovation but also multiply data security risks:

1. Massive Context Windows in AI Models—AI models will soon handle libraries of books, whole codebases, and multi-year datasets in a single query. This means AI will process more personal and proprietary data than ever before.
2. Self-Learning AI Agents That Operate Without Oversight – Instead of responding to prompts, AI will actively test, iterate, and improve its understanding without human oversight. This raises enormous ethical concerns when applied to sensitive financial or personal data.
3. Text-to-Action AI Will Automate Entire Industries – We are approaching a world where anyone can tell an AI, “Build me a new investment algorithm,” and it will instantly generate, refine, and deploy it. The power shift in financial and business strategy will be profound—but who controls the data powering these systems?

Put these together, and you have a future that dwarfs social media’s impact, turning every individual into a personal creator or innovator. However, the risks become astronomical if that future is built on stolen, scraped, or non-consensual data.

The AI Arms Race: Compute Power vs. Data Control

LeCun’s prediction that AI will move beyond LLMs is undoubtedly correct, but the real battle isn’t just about who has the most computing power—it’s about who controls the best training data.

China is aggressively building alternative AI ecosystems with:

• Domestic semiconductor development (Huawei’s Kirin 9000S, SMIC’s 7nm chips)
• RISC-V and open-source AI models to bypass Western-controlled architectures
• Federated learning and decentralized AI training to reduce reliance on cloud computing
• Massive national AI clusters and BRICS partnerships to develop alternative AI infrastructures

Meanwhile, the U.S. is ensuring that next-generation AI architectures remain within its cloud ecosystem—AWS, Azure, and Google Cloud—solidifying its dominance. However, if the U.S. wants to maintain AI leadership, controlling computing resources is only half the battle; the real power lies in controlling structured, high-fidelity data.

The Real AI Bottleneck Is Data Integrity, Not Computing Power

We assume that AI hallucinations, weak reasoning, and planning failures are hardware limitations, but they are data problems.

Clara Shih, VP of Business AI at Meta, reinforced this point in CEO Insights Asia:

“There’s no question we are in an AI and data revolution, which means we’re in a customer revolution and a business revolution. But it’s not as simple as taking your data and training a model. There are data security, access permissions, and sharing models that we have to honor.”

The next incredible AI leap will involve faster processors and structured, high-integrity data.

The Future of AI Will Be Decided by Data Control

LeCun is right—AI is about to undergo a seismic shift. But the question isn’t just how powerful AI will become; it’s who will own and control the data that fuels it.

We’re at a crossroads:

1. If data is monopolized or mismanaged, we risk an AI future controlled by a handful of players whose foundations are ethically dubious.
2. If we prioritize first-party, consented data and structured information, AI can become a tool for human progress—rather than an exploitative black box.

The Visa case and the LoanDepot breach are not just cautionary tales—they preview what happens when data control is unchecked. As AI accelerates and integrates deeper into critical infrastructure, including financial systems, business operations, healthcare systems, and national security, the risks of monopolization, exploitation, and breaches will only grow.

The real danger isn’t just losing control of AI—it’s allowing a select few to dictate its future. If we fail to rethink data governance now, we risk locking ourselves into a future where AI’s power is concentrated in those who control the data that fuels it. The time to act is before AI reaches its next paradigm, not after it’s too late.