Every generation has its “new hotness.” Fire. The wheel. Bronze. Iron. Electricity. Radio. The transistor. The internet. Blockchain. We’re wired to chase shiny things. Right now, that shiny thing is AI.
Everyone’s racing to build with it, integrate it, or slap it into their pitch deck. There’s an urgency in the air — figure out your AI strategy or get left behind.
But I think the spotlight is slightly off. AI might be the engine, but the real play? It’s in the fuel.
Data Is the New Petroleum
If AI is the internal combustion engine of our time, data is the fuel. And not just any data — clean, real-world, structured data. Without that, the most powerful models in the world are just math experiments.
History rhymes. When the car became mainstream, the big winners weren’t just car makers. The real leverage was in petroleum — extraction, refining, distribution. Fast forward to today and you see a familiar pattern.
The OpenAI and Anthropic types are absolutely going to print money. Jensen Huang called it: they’re becoming token factories — churning out intelligence one prompt at a time, each token a little drop of value. There’s no doubt about that.
But under all that are companies like Snowflake, Databricks, Scale AI, Palantir and LiveRamp. They’re quietly building the data supply chain. And in many ways, they’re sitting on the modern-day oil fields.
The Data Pollution Problem
Here’s the irony: these powerful models were trained by scraping the internet, but now they’re scraping their own outputs. That’s a dangerous feedback loop. We’re poisoning the well with synthetic noise and the signal is fading fast.
What’s left is raw, untapped real-world data. Human behavior. Environmental sensors. Financial flows. Traffic. Movement. Emotion. Energy usage. Video feeds. That’s the kind of data that can power the next wave of intelligence. But it’s not easy to get and it’s not evenly distributed — and fraught with privacy peril.
Some industries are handling this better than others. Healthcare, for example, has always had strict guardrails around data — but that’s also made it harder to train AI systems responsibly. One standout exception is Sirona Medical, which is working on responsible ways to apply AI to medical imaging.
Another example is Novi Connect, which gathers product information directly from brands to provide reliable, verifiable data. This kind of clean, first-party data helps ensure AI models are grounded in fact, not fiction.
The Missing Piece: Energy
There’s another layer here that doesn’t get enough attention.
None of this scales without energy. The sheer cost of running token factories, moving data, training new models — it’s massive. If we’re serious about long-term AI infrastructure, we need cleaner, more abundant energy.
If data is the petroleum, then energy is the steel.
Not just better GPUs. Better energy systems.
I’m keeping an eye on fusion, wave energy and other deep-tech solutions. Projects like Panthalassa are exploring the intersection of clean energy and planetary-scale infrastructure. That’s the kind of thinking that will actually support the AI era — not just for a few months of hype, but for decades of real progress.
My Take
This is just my personal opinion. Others may see it differently. But from where I sit, it looks like we’re watching history repeat itself.
AI will become normal. Expected. Boring, even. Just like electricity is today.
The real leverage is upstream — in the data and in the energy that powers it.
So while everyone’s caught up in what the models can do, I’m more interested in who’s building the pipelines, the refineries and the power grid underneath.
Because once the hype fades, that’s where the lasting value will live.