🧲 The first edition of Vertigo explained why I would write it during 1000 days: “Some say we are 3 years away from an artificial intelligence smarter than the smartest human on all levels. 1000 days might be what remains.”
Well 10 days ago Sam Altman wrote : “This may turn out to be the most consequential fact about all of history so far. It is possible that we will have superintelligence in a few thousand day”
This is PART 1 of a series that’s gonna dive into the superintelligence paradigm.
An absurdly complex topic. But you should grasp the stakes.
Menu of this edition
⌛VERTIGO OF THE WEEK - the rise of superintelligence: 4 things you will want to follow
💚 EMOHACKS - 4 ways to approach it simply
⚖️COACHING STORIES - A quick survey
🗞️ BREAKING NEWS / HOT LINKS - 2 must read and a vertigo video
⌛VERTIGO OF THE WEEK
You can’t escape it.
The great Nick Bostrom defines superintelligence as "any intellect that greatly exceeds the cognitive performance of humans in virtually all domains of interest" “including scientific creativity, general wisdom and social skills.” A new intelligent species basically.
Scientist Danny Hillis compares what’s happening to that point “when single-celled organisms were turning into multi-celled organisms. We are amoebas and we can’t figure out what the hell this thing is that we’re creating.”
“The first ultraintelligent machine is the last invention we’ll ever make. The last challenge we’ll ever face.” I. J. Good (1965)
3 examples of what this might do:
AI Automation: Full-fledged AI agents could automate software engineering and other intellectual jobs, effectively replacing human labor with autonomous digital workers.
Unstoppable Military Dominance: Superintelligent AI will grant overwhelming military advantages, potentially outpacing and taking control from pre-superintelligence systems.
Endless Innovation: Imagine the power of 100 million researchers working tirelessly day and night, accelerating discoveries and breakthroughs at an unimaginable pace.Gives you an idea?
You’ve just seen the tip of the iceberg so far: text, images, songs, videos, podcasts..
That’s generative AI: GAI. With a “G”.
Like in Gadget.
There’s an ongoing race for another G: General AI.
An AI as smart as a human.
And from there ASI, or superintelligence. Smarter. Way, way smarter. When things could look like this.
For AGI, it’s not so much about whether it will happen rathen than WHEN.
We’d like things to remain the same. They won’t.
What was a sci-fi thing is becoming increasingly real. Plausible.
So let’s approach it not as a potential sci-fi scenario but just what it is: data. Variables. Figures. We sort of know what we need to get there by now.
So what do we need to keep an eye on? It comes down to 4 acronyms.
1. GPU = COMPUTE
How do you go from there to there? You produce graphics processing units.
Picture for a second a 27*11cm card.
Half an A4 page basically.
But with 80 billion transistors on it. 80 000 000 000.
That’s the H100 GPU Chip: a $40,000 semiconductor marvel, involving nanometer-scale precision, heat dissipation, and energy efficiency.
In 2023, NVIDIA produced only 550,000 H100, with big tech companies racing to buy them.
B100, the next gen, is coming soon, and by 2025, the X100 will bring even more improvements. More. Better. Soon.
The point being, it moves fast.
In a decade, expect GPUs with 1 trillion transistors. 🚀

2. B$= MONEY
Chips do cost money, but so does the infrastructure behind & the power. The race for the most important cluster has started.
Follow the investments…
They have increased tremendously and total AI investment could be north of $1T annually by 2027.
Microsoft and OpenAI are rumored to be working on a $100B cluster, slated for 2028 (a cost comparable to the International Space Station!).
Last week news announced a massive $100B investment plan by Microsoft and BlackRock, as well as the announcement that OpenAI closed the largest VC fundraising round of all time at $6.5B.
3. GW= ENERGY
Following the current increasing trend, 6 years from now a trillion-dollar, 100GW cluster alone would require ~20% of current US electricity generation
But keeping it in the current “real” world, rumors suggest a 1GW cluster is being built in Kuwait. 1GW is the equivalent of 1,3 million horses strengh.

With multiple hundred MW installations in the work, the current datacenter consumption will undeniably increase its global power share.
💡 Don’t get me wrong, I am fully aware of the absurdity of this in an energy depletion paradigm. But that’s a story for another time….
Energy-rich countries will lead the race. Rings a bell?
4. FLOP = BRAIN
We can measure progress toward AGI by looking at FLOPs (floating-point operations per second) and compute power.
One way to estimate the brain's capacity is by calculating the total calculations per second (cps) it can manage. Legendary Ray Kurzweil used a shortcut by taking estimates for cps in certain brain structures, scaling them up based on brain weight, and arriving at a figure of 10^16 cps, or 10 quadrillion cps.
Ajeya Cotra offers a somehow similar estimate, suggesting that 10^18 FLOP—or one exaFLOP—is needed to simulate the key aspects of human intelligence. An “evolutionary bound”.
⚡The key takeaway is that AGI progress is measurable, and we're moving closer to that threshold.⚡
💡20 years: a 50 million-fold increase in computational power
In 2004, I was in charge of hardware for my business school video association. Back then, I used an NVIDIA GeForce4 visual card to process our projects. We did some fun stuff—3D emulation, After Effects work...really pushing the limits at that time.
Fast forward 20 years. The leap in technology is mind-blowing. We’ve gone from 20 GFLOPs (billions of operations per second) with that old GeForce4 to 1 exaFLOP (quintillions of operations per second) with today’s hardware, like the NVIDIA H100.
💚 EMOHACKS
What can you do about it?
Get informed: don’t miss on this one. Get used to it.
Last week I generated this amazing 3 minute podcast in 2 clicks.
This week I generated these photos through a Lora (Low-rank adaptation), a technique that simplifies complex machine learning models. It trained for less than 5 min on selfies that I gave it and can now generate a photo of me in any context..





Remember this astrophysicist flabbergasted by ChatGPT reproducing in 10 minutes a task that previously took him 10 months to code?
All this was unthinkable a few years back.
Think a few years from now.
Read. Listen. Get informed. Write me if you want advice on what to read and listen.
Accept that you might not get it.
We don’t have a word for an IQ of 12,952
💡 “Superintelligence of that magnitude is not something we can remotely grasp, any more than a bumblebee can wrap its head around Keynesian Economics. In our world, smart means a 130 IQ and stupid means an 85 IQ.
We don’t have a word for an IQ of 12,952.” tim urban
Keep cool.
It’s gonna change every month.
And if you’ve read correctly, while the H100's theoretical exaflop technically puts us close to human intelligence, we're not there yet.
AGI requires more than just raw computational speed; it needs precise neural architecture, flexible cognitive abilities, and complex learning mechanisms beyond what's currently possible with today's hardware.
Reconnect with non-digital parts of your life.
Well that’s actually one I recommend beyond AI,AGI, ASI, GenAI…
Read these 4 simple hacks to harness nature power in Vertigo#6
⚖️COACHING STORIES
It’s a dense one already so I’ll skip that part.
I’m working generally around anxiety and how acceleration exposes us all.
The near future will require that we navigate this anxiety in its different forms.
🗞️BREAKING NEWS / HOT LINKS
To prepare this edition I went through the Situational Awareness report from Leopold Aschenbrenner, a brillant overview of what’s at stakes, and a deep insight of the Silicon Valley mindset.
Long before that, and in a more accessible way (as always), Tim Urban wrote this now classic piece on AI. A great, fun and well-put introduction: part 3 is now long overdue Tim!
And watch this amazing 2 minute extract from the no-less amazing movie EX MACHINA. Please share your movies and books about the topic!
PS: Have you taken this 8 question test? You’ll get a free personnalized feedback?
Or even better, why don’t we have a chat? I’d love for you to be listened to like never before.
Last thing, I would not want you to leave without a good playlist you can listen to while reading or just chilling.