See All Articles on AI
Every week in tech feels big. But some weeks feel like the world is quietly rearranging itself while most people are still trying to catch up. This was one of those weeks.
From Anthropic overtaking OpenAI in enterprise adoption, to China shattering drone world records, to Europe scrambling to stay relevant in the AI race — the pace of change is dizzying. And threaded through all of it is a deeper tension: the world is hurtling toward superintelligence and abundance, while millions of people are still worried about paying the rent.
Here’s the breakdown of what really mattered, why it matters, and what’s coming next.
1. Anthropic vs. OpenAI: The Silent Battle for the Future of Intelligence
For months the narrative has been dominated by OpenAI. But in the enterprise LLM market, something surprising happened — Anthropic overtook OpenAI in API adoption.
This isn’t just a chart on Hacker News. It's a clue to a deeper philosophical split in the AI world:
-
OpenAI is betting on multimodality — video, image, audio, simulation.
-
Anthropic is betting on code — building models optimized for software generation and recursive self-improvement.
If code is the key to AGI, Anthropic may be quietly building the stronger long-term position. If the “special sauce” lies elsewhere, OpenAI’s broader model capabilities may win out.
But here’s the bigger shock:
Anthropic expects $70B in revenue by 2028 with 77% profit margins, while OpenAI expects $100B by 2029 — but still unprofitable due to capital expenditure.
Welcome to the era where LLMs become trillion-dollar utilities.
2. World Models and the Coming Holodeck Wars
Fei-Fei Li’s new company, World Labs, revealed something jaw-dropping: a model that generates entire 3D worlds — not pixels, but fully traversable environments built from millions of Gaussian splats.
Imagine:
-
AI agents trained inside synthetic universes
-
Games created instantly from text
-
Photorealistic VR worlds that feel indistinguishable from life
-
AI-powered “holodecks” as a platform, not a fantasy
This is not entertainment technology — this is infrastructure for future intelligence.
The biggest market here isn’t gaming. It’s synthetic data and robotic training that could replace thousands of real-world experiments.
We are at the beginning of the Holodeck Wars, and the implications are outrageous.
3. AI That Can Forget: The Rise of Machine Neuroplasticity
A breakthrough paper introduced a technique that lets AI models forget memorized private data without losing reasoning ability.
Why this matters:
-
It enables smaller, more efficient models
-
It reduces hallucinations tied to memorized facts
-
It supports enterprise privacy
-
It moves us closer to micro-models with <1B parameters that still perform like giants
This is machine neuroplasticity — pruning the brain while keeping the intelligence.
If this trend continues, the next frontier models may not need trillion-parameter behemoths at all.
4. China’s Open-Source Shockwave: $5M for a Trillion-Parameter Model
The most under-reported story may be the most transformational:
Moonshot AI (backed by Alibaba) released an ultra-low-cost open-source model that runs on Groq hardware and competes with top Western models.
Training cost?
$4.6 million.
This breaks the capital advantage of U.S. AI giants. If anyone can train a frontier-class model for under $5M, the competitive map changes overnight.
This is the moment “AI for the few” becomes “AI for everyone.”
5. Europe Loosens GDPR — Too Little, Too Late?
After years of regulatory paralysis, Brussels is finally softening GDPR restrictions to allow AI innovation.
Why?
Because Europe woke up to the reality that:
-
AI startups are launching 6–12 months later than U.S. competitors
-
Venture funding dropped 30%
-
Mandatory AI audits cost €260,000 and take up to 15 months
-
Talent is fleeing to the U.S. and Asia
Europe’s dream of “ethics first” collided with economic gravity.
But is this change enough? Only if Europe can simultaneously:
-
speed up energy expansion
-
build data infrastructure
-
remove bureaucratic sand
-
and retain talent
The window is closing fast.
6. The Real Global Crisis: People Are Scared
Across 32 countries and 60,000 respondents, the top three concerns were:
-
Cost of living
-
Unemployment
-
Inequality
People aren’t thinking about AGI. They're thinking about survival.
We talk about exponential abundance — and it is coming — but not fast enough for the billions who are hurting. The next 2–7 years will be turbulent. Jobs will be displaced before economic systems adapt.
If people don’t believe in a hopeful future, fear narratives win. And fear is the oxygen of backlash.
This is the real challenge of the AI era:
accelerate abundance without breaking society in the process.
7. Data Centers, Energy, and the Coming Power Crunch
Here’s a stat that should terrify world governments:
The U.S. alone will need 92 gigawatts of new power for AI by 2030.
New nuclear reactors (AP1000 class) take 5–10 years to build.
Even with an $80B nuclear restart plan, we’re still woefully behind.
If we don’t fix energy:
-
AI stagnates
-
The economy stagnates
-
National security collapses
-
And superintelligence becomes impossible
Energy is the single scarcest resource for the future.
8. Swarm Robotics: The New Infrastructure of the Physical World
China coordinated 16,000 drones with millimeter precision — the largest controlled swarm in history.
This isn’t a light show.
It’s the beginning of a new physical platform.
Drone swarms will:
-
construct buildings
-
fight wars
-
deliver goods
-
clean cities
-
repair infrastructure
-
act as distributed sensor networks
Humanoid robots get the headlines.
Swarms will do the heavy lifting.
9. The Tesla Flying Car Might Actually Be Real
Elon Musk hinted the next Tesla Roadster may include cold-gas thrusters from SpaceX — yes, actual rocket tech.
Not to fly like a plane, but to:
-
hover briefly
-
leap over obstacles
-
accelerate violently
-
reduce crash impact
It sounds insane.
Which is why it’s probably happening.
This is classic Elon: replacing marketing spend with audacity.
10. Geoengineering Goes Mainstream
Elon floated an idea: a solar-powered satellite constellation that can regulate Earth’s temperature by adjusting how much sunlight reaches the planet.
A global thermostat.
Science fiction?
Not anymore.
This is reversible geoengineering — the safest version we have. But it’s also a political minefield. Some countries want warming. Others are drowning because of it.
Still: without geoengineering, climate timelines don’t work.
11. Blue Origin Finally Sticks the Landing
For the first time, Blue Origin successfully landed its massive New Glenn booster. This gives humanity a second reusable heavy-lift path to space.
This matters because:
-
SpaceX can’t carry the entire planet’s ambitions
-
Competition drives innovation
-
Orbital logistics become more resilient
-
Starlink finally gets a competitor
We’re entering the era of commercial railroads to orbit.
12. The Backlash Begins: Boston’s Unions Fight Waymo
Boston labor unions formed a coalition to block driverless cars unless they include a human driver — which defeats the point.
This is not about safety.
This is about job loss anxiety.
Expect more of this globally.
People are terrified, and fear fights innovation.
Unless we build social cohesion tech — policies, safety nets, narratives, and new economic models — this friction will escalate.
The Big Question: What Future Do We Believe In?
Technology is accelerating 40x year-over-year in capability and cost deflation. But humanity isn’t accelerating with it.
We have two paths ahead:
A world where abundance rises and lifts everyone
— energy becomes cheap
— healthcare becomes free
— education becomes personalized
— AI agents make livelihoods easier
— global prosperity expands
A world where abundance rises but only for a few
— fear spreads
— inequality widens
— social unrest grows
— innovation slows under backlash
— and opportunity shrinks
Which future we get depends on how we handle the next few years — not the next few decades.
This is the decade the world remakes itself.
