The wall between imagining software and making it exist just collapsed. For decades, the limiting bottleneck was syntax. People had clear, painful visions for things that ought to exist, and were stopped dead by the compiler. That barrier is gone. It feels miraculous, and we have the strange privilege of standing inside the shift.
It also arrives with an unspoken debt.
We speak about artificial intelligence as though it floats weightlessly in the ether. It does not. Silicon runs on substations, cooling towers, and diverted river water. We are transacting directly in joules.
The Physics of a Prompt
A frontier model returning a breezy “hello” consumes the watt-hours that could have kept an LED reading lamp burning in your room for minutes.* Silicon compute is physical work. Soon, people will transact directly in tokens: compute will be the reserve currency of the intelligent economy, and energy will be the underlying peg.
*Google published 0.24 Wh as the median Gemini text prompt (Aug 2025) — roughly three minutes of a standard 5 W LED reading lamp.
Which exposes an uncomfortable truth: people who lecture about ESG, sustainability targets, and energy conservation while aimlessly burning AI tokens are being hypocrites. Firing off millions of automated inference cycles to produce corporate sludge, bloated slide decks, and speculative noise is not innovation; it is digital pollution.
The Antidote: Absolute Intention
If every prompt incurs a thermodynamic withdrawal from the electrical grid, the only moral defense for using this much power is intention: building software that genuinely lightens the friction of someone’s real, physical day.
- My wife is a pharmacist: so I spent energy building PharmaRab and Provia to help her and her colleagues conquer grueling licensing exams.
- My friends are physiotherapists: so I spent energy building PhysioArab and kinē to build clinical rehabilitation tools that therapists needed.
- Dubai’s finest cooking was hidden next door: so I built NxtDoorChef to connect neighbors directly to authentic home kitchens.
- I collect mental models: so I built Doomscholar to map frameworks that survive the hype cycles.
The True Moat
Tech is never the moat. What it alters in someone’s real, physical day is. Nobody values how convoluted your build pipeline was. They only value whether it helped.
Around a full-time job and family in Dubai, I don’t just read release notes. I test new tools by building tools that work. This domain is where they live: an open ledger of software built in public.