In 2021, AI cracked protein folding, a breakthrough destined to save millions of lives. Since then, we've mostly taught robots to write emails nobody wants to read.
We possess computational power that could reshape civilization. Yet we use it to generate anime girlfriends and automate rejection letters.
U.S. data centers already consume 4.4% of national electricity, with that number projected to nearly triple to 12% by 2028. But the real tragedy isn’t the electricity itself, it’s how we squander that immense power.
The Reality Check
The internet was supposed to democratize information. Instead, we got Facebook uncles and conspiracy theories.
AI’s following the same script. Data centers are racing toward double-digit shares of America's electricity consumption, with AI feeding most of that surge.
The 1% Doing Real Work:
UNHCR’s Project Jetson predicts refugee movements, improving humanitarian response planning.
GiveDirectly targets aid via satellite, reaching families in crisis more efficiently.
AI models track disease outbreaks in refugee camps to help stop the spread.
Drug discovery timelines dropped from 10 years to 3 for certain compounds.
Climate models are now accurate within 2°C for 5-year projections.
Latimer.ai tackles AI bias by weaving diverse cultural perspectives into training data — fighting erasure that affects billions.
The 99% Burning Compute:
ChatGPT homework cheating (>75 billion queries monthly)
Midjourney profile pics for crypto bros
Generative AI spewing billions of spam emails daily
Virtual girlfriends
SEO content farms planting nothing but slop
GitHub Copilot enabling copy-paste coding
Stable Diffusion creating non-consensual imagery
The numbers tell the story: for every model predicting droughts, there are 10,000 churning out LinkedIn posts about “synergy.”
The Arithmetic of Waste
Every GPU cluster minting NFTs could instead model insulin distribution across sub-Saharan Africa. Every Bitcoin mining operation (110 TWh annually) could optimize vaccine delivery worldwide. Every chatbot training session could tackle antibiotic resistance.
The economics are ruthless. Problems lacking subscription tiers don’t get solved.
Malaria kills 600,000 annually but the cure has no SaaS subscription. Preventing famine doesn’t deliver quarterly growth. No one’s raising series B to cure dysentery.
So, we get AI Drake songs, automated harassment bots, and “productivity” apps accelerating busywork while millions die preventable deaths.
Silicon Valley’s Investment Thesis:
Where the money flows:
Autonomous vehicles: A $68B annual market (because sharing the road is déclassé).
AGI research: $100B+ sunk cumulatively into top labs training our eventual robot overlords.
Crypto/Web3: Market cap hovering over $62B (digital tulips galore).
Space ventures: $6.6B in annual deals (the ultimate white flight).
The metaverse: A $154B global market (where your avatar can also rock a fleece vest).
VR/AR: $46B-$100B market size (perfect for ignoring the reality we’re creating).
What can’t get funded?
Malaria prevention: A $5.3B annual shortfall while 600,000 (mostly kids) die.
Clean water infrastructure: $131.4B funding gap leaves 2.2 billion parched.
Soil depletion research: $278B annual deficit (Mars soil will probably be fine).
Antibiotic resistance: Just $66M new funding in 2024 (against 10M deaths projected by 2050).
Climate refugee support: A cumulative $28M from Global Cities Fund (doesn't even cover relocation consulting fees).
Capital flows exclusively toward problems that fascinate the Patagonia vest demographic.
The Dependency Trap
What we dub “progress” is subscription-model dependency. Job applications need AI-written cover letters. Students can’t compete without ChatGPT. Every pro tool flaunts an “AI assistant” you can’t turn off.
We didn’t adopt this tech. It adopted us. Silicon Valley built the cage; we paid to be locked inside.
Available Futures We’re Ignoring
Today’s tech, not theory or wishful thinking, enables:
Crop failure prediction 6 months ahead (potentially preventing famine for millions)
Medicine routing algorithms (30% fewer deaths from shortages)
Heat-resistant city modeling (saving 100,000+ lives annually)
Economic policy testing before the chaos hits
Real renewable grid optimization (40% efficiency gains possible)
Illegal fishing tracking (preserving food security for 500M)
Pandemic early warning systems (14-day advance notice)
These aren’t fantasies. They’re sitting in GitHub repos waiting for compute time.
The Call
Engineers: Your genius is optimizing ad clicks while the world burns. Demand 20% of compute be devoted to problems that matter. Push for public GPU allocation dashboards. Make the waste visible.
Investors: Stop fixating on TAM. Start asking about lives saved. Fund drought prediction, disease modeling, refugee logistics. Accept 5% returns that include survival.
Companies: “Google Prevents Famine” beats any earnings call. “Microsoft Solves Drought” outvalues your market cap. Choose legacy over liquidity.
Everyone: Support transparency laws on compute and energy use. Demand AI not wasteful but life-saving. Vote for leaders who get that innovation without survival is suicide.
The Reckoning
This isn’t anti-tech. It’s anti-waste.
Big tech perfectly models climate catastrophe internally. They choose engagement metrics instead. They’re betting the cash-out timeline beats the consequence timeline.
In 2040, your kids will ask what you did when humanity gained godlike power. “Made emails 10% faster” won’t cut it.
We have less than a decade to steer this tech from entertainment to survival. The choice isn’t AI or not; it’s distraction or salvation.
The clock is ticking. History won’t forgive wasted power.
Share this with engineers optimizing the wrong metrics. Send it to investors confusing disruption with progress. Forward it to anyone who thinks Silicon Valley’s incentives align with humanity’s needs.