Last week at Cota Connect, I had the privilege of hearing Bill Gross (legendary entrepreneur and founder of Idealab) give a talk he called Game Changers.
He traced his entrepreneurial journey, from building speakers in his dorm room, to the earliest days of Idealab, through to the profound transformations unfolding in the AI era today.
One idea stayed with me after I left the room. Bear with me…
Chlorophyll runs at about 1.5% efficiency.
That is the real-world figure: the percentage of incident sunlight that a plant actually converts to biomass over a growing season. The rest is reflected or lost as heat.
That has been the engine of all life on Earth, including every civilization humans have ever built.
The pyramids. The Silk Road. The Roman aqueducts. All of it ran on converted sunlight. Sun to leaf. Leaf to grain. Grain to calorie. Calorie to muscle. Muscle to stone. Stone to civilization.
1.5% efficiency. And it built everything.
Modern solar panels convert sunlight at roughly 22%.
Nearly fifteen times the efficiency of a leaf. It took four billion years of evolution and about a century of materials science to get there.
The energy capture problem, the one biological life spent four billion years working on, is on its way to being solved.
The Industrial Revolution broke the dependence on biological conversion entirely. We stopped feeding energy into stomachs and started feeding it into machines. Steam engines, turbines, assembly lines.
Energy powered muscle. Then mechanical force. Now intelligence.
As Bill Gross pointed out, we feed watts into a data center and out comes a token, a unit of thought. We are, quite literally, turning watts into wisdom.
The appetite for watts is becoming insatiable.
According to IEA estimates, a single AI query consumes roughly ten times the energy of a standard web search. Multiply that by billions of queries a day, by the training runs that precede them, by the infrastructure that supports them, and the energy demands of the AI era are unlike anything the grid was built to handle.
Solar, wind, and every available clean energy source are being deployed at a pace we have never seen before, not primarily for homes or cities, but to power intelligence.
The AI era doesn't just use energy differently. It devours it.
We know how to frame that problem. It is an infrastructure challenge, hard, expensive, urgent, but legible. We can measure watts. We can build capacity.
The harder problem sits at the other end of the chain.
Closed, it is the most valuable problem being worked on right now. Open, it is the most expensive one.
What is the output of this energy worth?
At every stage of human history, this question has been difficult. But it has always been answerable.
The farmer knew what her grain was worth, the market told her, in weights and prices.
The industrialist knew what his output was worth, a cost per unit, a margin per sale.
Value was tethered to something physical. Something you could count, weigh, or exchange.
Intelligence breaks that framework.
Not because the value isn't there. It clearly is. A decision made better. A problem solved faster. A risk identified before it compounds. An opportunity seen before the competition sees it. These things have real economic consequences — sometimes enormous ones.
But intelligence is unlike every previous unit of value in the chain:
Its value is realized in the future, not at the moment of production. It is context-dependent, the same insight is worth very different amounts to different recipients. Its contribution to any outcome is rarely clean: did the deal close because of the analysis, the relationship, or the timing? And it has no physical proxy. Every previous era had one, calories, tonnes of steel, kilowatt-hours.
Tokens are a unit of computational work. Not a unit of delivered value.
The chain produces them. It does not yet know what they are worth.
Every era improved on conversion efficiency.
Chlorophyll to solar panel: 1.5% to 22%.
Human muscle to mechanical force: orders of magnitude.
Computation to intelligence: extraordinary.
Intelligence to proven value: still unsolved.
The hardest problem, the one that has persisted across every era, was never on the supply side. It was always on the demand side.
What did the output actually deliver? To whom? And how much is that worth?
The farmer answered it with a scale. The industrialist answered it with a margin. The AI era has not yet answered it.
Chlorophyll ran at 1.5% efficiency and built everything anyway. Every era solved its measurement problem with imperfect tools, under pressure, because the economy left no other choice.
The AI era will be no different.
What stays with me — what has stayed with me since I left that room — is not the elegance of the chain from watts to wisdom. It is the size of the gap at the end of it.
Large language models. AI assistants — ChatGPT, Claude, Gemini. AI agents operating autonomously across entire workflows. AI platforms that whole businesses are being rebuilt on. AI applications growing at rates that make the early internet look gradual. Every category expanding faster than the last, each one generating intelligence at a volume no previous technology has approached.
The infrastructure to measure what that intelligence delivered has not kept up.
Closed, that gap is the most valuable problem being worked on right now. Open, it is the most expensive one.
The gap at the end of the chain is exactly the problem valueIQ was built to close. Try it free and turn the intelligence you deliver into a value case your buyers can actually see.
Amar Dhaliwal is CEO & Co-Founder of valueIQ.
