Software engineering is about complex systems.
However not all software is complex. Lot of software is very simple.
AI will replace software engineers developing those simple systems - if not replace, it will shrink the demand for them.
1
1
12
If the software to build starts to get complex - maybe AI is not mature yet for developing software with high scalability or more advanced use cases
1
6
Writing specs or managing agents is the new coding.
Specs and skills became portable - this is the new code.
7
Giovanni Gaglione retweeted
Infrastructure Expenditure by Country
China spends 4.8% of their GDP on infrastructure, America spends 0.51% of GDP on infrastructure.
What is infrastructure?
It’s the grid, power generation, transport, logistics, airports and shipping terminals, rail and roads, the water and sewerage.
According to Cato Institute all US infrastructure public + private is worth $28-32 trillion.
cato.org/sites/cato.org/file….
According to IMF studies all China’s infrastructure is worth $32-33 trillion.
researchgate.net/publication…
The two superpowers are basically at parity in terms of their infrastructure stock.
HOWEVER, China is spending 10x what the US spends on building and maintaining their infrastructure.
Now the second law of thermodynamics is the same on both sides of the Pacific Ocean. Entropy degrades everyone’s infrastructure over time at crudely the same rate (big simplification but it stands).
BEA puts the capital weighted age of all US assets at 29.5 years, up from 22.1 years in 2000.
CFA puts the capital weighted average age of all Chinese infrastructure assets at 11.8 years.
Consider how long running and durable these trends are.
China consumes 46.3% of all the world’s steel. USA consumes 5.3% of all steel. Again it is 10:1 it’s all flowing into infrastructure.
What are they building in China?
China is a machine.
I mean that in a real way, you should think of China as a machine that is engaged in recursive self improvement.
You can think of any country as a machine (or a system). But the map below shows that one of them is clearly doing something unusual.
Infrastructure is cumulative. The average capital age and the percentage of GDP spending can tell you how much infrastructure any country will have each year, long into the future.
Economics is soon going to uncouple from demographics, as AI and automation grow faster than the incumbent economy continues to grow and overtake it.
What limits the speed of economic growth once GDP is free of demographic limits?
Yup, infrastructure.
The world has to build a lot of infrastructure.
Any company should turn into an AI-native company.
At work, any process and activity should be supported by an agent harness.
Knowledge, experience and agent automation should be on a constantly-improving loop.
This is key nowadays to avoid digital divide
9
For software engineering activities of course I use agents.
I'm happy that I moved my agents from working on 3 minutes tasks to 25 minutes tasks.
It's not only about the time: I'm delegating more to the agents, the work is much more structured and their output is higher quality
1
54
This weekend, in Florence, I saw the painting "Madonna col bambino e angeli" by Botticelli.
Then I imagined how that could be pictured nowadays.
Here it is - both side by side.
15
Where the value is going nowadays?
1) in scarcity
2) in automation
3) in 1000x productivity value creation
Own scarcity. Enable automation. Capture productivity.
2
17
AlphaEarth: 10m pixels, multi modal for a new dense and more efficient representation of the physical world
medium.com/google-earth/ai-p…
53
The market is no longer sure the annuity is an annuity.
As of early 2026, software trades at roughly 22.7x forward earnings, below the S&P 500 for the first time ever
saastr.com/the-sub-5-club-th…
1
64
Government should move from 'governing' to 'reviewing their Government agent'.
Voters should move from 'voting' to 'reviewing their voter agent'.
If so, I expect more effective and efficient states as outcome
#government #ai @elonmusk #postai
13
Post AI era ->
AI creates a world where value is produced faster than it can be distributed.
How to adapt distribution to the new value creation era?
1
14
How to reach your target audience and stand out to Llms
producthunt.com/stories/how-…
11



