Data Is Not the "New Oil": Where the Comparison Holds and Where It Misleads

Burn a barrel of oil and it is gone. A record of what millions of people have bought, by contrast, can train a model today and serve another tomorrow without losing anything. That is the first thing to expose the limits of calling data "the new oil", a comparison that caught on in the business press because it captures a truth: whoever holds the most valuable resource holds influence. But a metaphor lights up one side and darkens another.

Where does the comparison mislead?

First, data is nonrival. The economists Charles Jones and Christopher Tonetti describe data as nonrival: the same data can be used by many firms at once. This yields increasing returns, and making data available to a wide range of parties may bring social benefit even after privacy concerns are counted, whereas firms may withhold it for fear of competition and so waste its benefit. That logic looks nothing like the scarcity of oil.

Second, raw data has no value. Data that is not cleaned, tied to its context and turned into decisions is a burden to store and secure. It can pile up in scattered repositories that do not talk to each other. The engineering adage "garbage in, garbage out" applies: a model knows of the world only what it has been shown.

Third, it has an energy cost. Data centres used about 1.5% of the world's electricity in 2024, by the International Energy Agency's estimate, and the agency expects large growth (the figures are in the article "The AI Race: Who Owns the Chips, the Energy and the Rules?"). The digital fuel itself burns physical fuel, and any data strategy that ignores this is incomplete.

And where does it hold?

It holds in that the ability to make use of data produces an advantage. In a study of 179 large firms, those relying on data-driven decision-making were about 5 to 6% more productive than expected given their other investments and their use of information technology, and the researchers tried statistical tools that reduce the chance that the explanation is simply that such firms were better performers to begin with. The study is fifteen years old and does not concern generative AI, but it supports the idea that a culture that makes data the basis of decisions gains.

Accumulation is likely to lean towards concentration: whoever has more data improves their service, attracts more users and so gains more data. This is a plausible mechanism for which we have no figure here, so we present it as a hypothesis. Nick Couldry and Ulises Mejias argued in "The Costs of Connection" (2019) that extracting data from people's lives extends older patterns of extraction, concentrating power in a few companies. It is a contested thesis, which some researchers accept and others dispute; whatever one makes of it, it is a reminder that those who lack sovereignty over their data and models depend on others for their decisions.

Data is the trace of people

What the oil metaphor misses is that every data point is the trace of a person with privacy and dignity. That is why data is not treated as a pure commodity. In the Kingdom, the Saudi Data and Artificial Intelligence Authority (SDAIA) was established by royal order in 2019 and handles data-governance policy and supports implementation of the Personal Data Protection Law, according to a summary by a non-governmental body, so the official text should be consulted for details. In the balance of the Sharia, what we have been entrusted with of people's private matters is a trust to be safeguarded. On the cost of "consent" and the right of access, see the article "Is Consent Enough? The Cost of Reading and the Right of Access to Data".

Questions for any institution

1. What data do we really collect, and for what specific purpose?
2. Who has access to it, and for how long do we keep it?
3. Is it tied to its context so that it can support a decision, or are we piling up scattered files?
4. Do we keep the trust of the people it comes from, a capital that no quantity of data can replace?

The most useful thing to say may be to leave the question "how much do we collect?" and ask "what do we make of what we have, and on what basis do we protect it?" Whoever answers well makes data a source of trust and sound decisions, not a load that weighs down its owner.