AI is rapidly becoming a commodity, similar to electricity, with models converging in performance and racing toward lower prices. Infrastructure will remain essential but low margin, like the power grid. The real value will shift upward to the platforms that turn raw intelligence into usable productivity. The winners will be the horizontal work environments where all work happens, where AI, workers, and supervisors operate together, and where model choice becomes irrelevant because the platform orchestrates everything.

---

Every few weeks, experts debate whether AI is in a bubble. The arguments swing between infrastructure saturation, runaway debt loads, collapsing model prices, and the Manhattan Project mentality of spending whatever it takes. But while everyone obsesses over model benchmarks and GPU supply, a deeper shift is unfolding. AI is becoming a commodity, a raw input, an undifferentiated energy source. The real competitive frontier will not be in model creation at all. It will be in the systems that convert this energy into something users can actually feel.

To understand this shift, it helps to compare today’s AI landscape to the structure of an electrical grid.

AI As the Commodity Layer

AI models are ultimately becoming the electricity of the digital economy. They are powerful, essential, and increasingly interchangeable. As costs fall and model performance converges across providers, the source matters less than the availability. Once every company can access near equivalent cognitive power at near equivalent cost, the models stop being a moat. They become a public utility.

Model providers are racing to push out faster, cheaper, more efficient models, and that race will only accelerate. The price of inference will drop. The number of models on the market will explode. What was once proprietary becomes ubiquitous. At that point, AI is not the differentiator. It is the baseline substrate for everything else.

Infrastructure As the Grid

If models are the electricity, then the GPU clusters, cloud providers, and hardware infrastructures are the grid. Necessary, expensive, and invisible. Someone must build and maintain the grid, but the wealth does not concentrate there. The grid enables value but does not define it. That is why debates about GPU oversupply, trillion dollar buildouts financed at double digit interest rates, or whether Nvidia becomes the next Cisco all share a common flaw. They focus too much on the grid itself, not the world that will run on it.

Value moves upward, not downward.

Platforms As the Energy Translators

Once electricity is abundant, the frontier shifts to the devices, appliances, and systems that turn that raw energy into human experience. In AI, those devices are the platforms that sit directly in the user’s workflow. These are the layers where human effort is removed, where work happens, where teams operate, where decisions are made, and where productivity gains are realized.

This is where differentiation and defensibility will live. Not in the models. Not in the hardware. Not in the training runs. In the platforms that orchestrate work horizontally across the entire organization.

Horizontal Work Platforms Are the Real Destination

The future will not belong to vertical SaaS tools with a sprinkle of AI. It will belong to the massively horizontal work platforms that handle messaging, tasks, decisions, oversight, alignment, reporting, and execution in one integrated environment, then rapidly train up synthetic agents to take over the work. These platforms will plug into every model the world produces and will dynamically choose the best one for the job. As model performance converges and cost collapses, the advantage shifts to whoever controls the place where work is actually done.

That is the environment I have been building toward for more than two decades. A single plane where workers, supervisors, and AI all operate together. A place where the user does not care which model provided the insight, only that the insight was delivered at the right moment with the least possible effort. A place where autonomy becomes accessible.

As AI continues to commoditize, the companies that win will be the ones that turn this new electricity into human productivity at scale. The grid will matter. The models will matter. But the real value will accrue to the platforms that put all of that power directly in the hands of their human operators with the least amount of friction.

…