lnkdinsuxs
OpenAI has been gearing up for RSI. Jakub Pachocki has mentioned this as a priority, even as recently as September 6 in public. GPT-5.3-Codex helped with its own development.
They have been experimenting with it for post training (GPT-Red is a documented model-to-model training loop) but it's also likely there's pre training for it.
What no one is talking about so far is how Intel used to be a strong engineering culture but then turned into an incestious den of bean counting MBAs executing HBS management best practices to maximize shareholder value.
Intel just doesn't pay their engineers enough, so most don't care. Just like most of all the large companies - to management, engineers are line cooks flipping burgers at McDonalds who just happen to have master's degrees.
Intel's decline is organizational - it allowed bean counting and management rituals to acquire too much influence over engineering judgment. Horrible punishment lies ahead - we need to track people in management who will now start to jump ship and take their style to other unsuspecting companies.
As a result of all this nonsense, it lost significant engineering talent to competing chip companies: Jim Keller went to Tenstorrent, lead GPU architect David Blythe joined AMD, and Xeon architect Sailesh Kottapalli moved to Qualcomm after 28 years at Intel. These are substantial transfer of expertise, whatever each person’s individual reasons for leaving.
In April 2025, Lip-Bu Tan, lamented the company was "too slow, too complex and too set in our ways". He pointed to teams buried under at least eight organizational layers, managers measured by their headcount, and excessive meetings and administrative work.
The financial priorities shifted: Intel spent $14.2 billion on share buybacks in 2020, the same year it disclosed delays to its 7nm CPU roadmap.
Now they pay the price as they confront the reality of an organization overloaded with TPS reports. The death will be slow and gradual, bleeding out slowly but surely - halal style.
AI infrastructure is supply constrained, so you have to make a decision: how much do I allocate to training vs inference?
Do I use it all for inference and do no further training? If so, what risks does that expose me to?
The U.S. is having its midterms - it closes tomorrow - and the next two years are pivotal to who wins the senate for the next 4 years. Having it leaked that billions of dollars of domestic AI spend is going to India is going to sink the GOP.
OpenAI has large offices in Delhi, Mumbai, and Bengaluru. Anthropic is rapidly expanding their Bengaluru office to Mumbai.
India is an extremely large market - there are 4x more people in India than in the U.S, so it's a huge TAM but most of them don't read or speak English so there's a tremendous localization challenge.
India is undergoing a fundamental shift in their economy. They used to be a culture that used to save money before they spend it. Recently I saw a large group of Indian visitors by the Golden Bridge taking pictures and asked them how they could have afforded to save for the trip given the current explosion in prices.
They hadn't saved money - they had taken out loans for the trip. Everyone of them. Apparently their house help earns as much as a teacher in Nonalva, Mumbai (I might have got the city wrong but it sounded like that)
They also focus their students on STEM and are having a tech. hiring crisis (so talent is really cheap). Every engineer speaks atleast 3 different languages, including English and even more local dialects.
Adjusted for cost, the above average Indian engineer assisted by Claude is insanely more valuable than an average engineer from SF assisted by Claude. It's like DeepSeek V4.1 Flash 1017 vs your average US frontier model.
As I said in another comment, $180k/yr doesn't go very far in San Francisco specially if you want to have a family but I won't be surprised if you could retire in India after working on $180k/yr for 2-3 years.
Every market has an upper threshold of costs parties are willing to bear and San Francisco passed it where there's now active demand collapse. Companies are voting with their wallets and people in San Francisco are paying the price.
because for entire companies, due to outsourcing, there is no other team to delegate to?
Who's the engg. team who writes the website for Best Buy, Target, eBay, Paypal?
I linked to the article that proves there's a path to replace surgeons. I don't see any other way to interpret that article.
> Why do you think entrepreneurship will be spared? Funny how you dunk on surgeons but then claim that you yourself are "protected" in the same breath.
Because entrepreneurship cannot be done by sitting in a room typing away at a keyboard.
You have to go out, talk to people but more importantly observe them, listen to them lie, tease out what they are actually willing to pay for from what they claim they will. Then you have to talk to investors.
Doctors used to be similar - they would listen to patients lie (most of the times unknowingly) and then ask probing questions. In 2026, no doctor who works at a hospital is allowed to do that. The hospital schedules their day back to back, they get to spend maybe 5-10m max per patient and figure out what tests to order and how to triage their concerns. All conversations are recorded and transcribed, fed to AI for training.
What used to be one visit with a doctor for 30-40m is now 10 visits, each 5-10m max scheduled back to back. Hospitals are working on replacing doctors with AI so those fat salaries can go to the bottomline.
At the scale where hiring overseas workers make sense, there are atleast one PM, QA people, Ops - likely teams, not just one person per, involved. No one is releasing a feature to prod by lunch.
The workflow you propose is a very good representation of SMBs and they are behaving as you have proposed - hire one or two local devs from Austin, call them over Slack and release to prod by lunch.
So yes, this is happening, just wanted to point out that the "overseas workers" workflow is at a different scale of operations.
How are you getting these numbers?
My feel is it's the opposite of a "detune".
Sol 6.1 is likely a distillation that's highly instruction tuned and designed to perform better on benchmarks.
On specific coding skewed benchmarks yes.
But Astra is in a different league when it comes to doing research and investigative tasks that require exploration.
For my usecases, Astra is pareto frontier - it dominates Fable 5.1 on cost. Fable 5.1 is a phenomenal model but Astra is cheaper and uses way less tokens and steps to produce similar output.
I agree that the H1B train is a red herring but I explained in my comment how AI has enabled mediocre talent to become more effective. Effective enough where they are "good enough" for what most companies need.
One of the reasons why offshoring was difficult was that asynchronous work wasn't the normal, so what a lot of companies would do is give the uninteresting pieces to the offshored teams.
That's taken a 180 in 2026. Google, OpenAI, Meta are building core infrastructure and software out of India. They don't even have to pay H1B rates.
Tech was already moving towards asynchronous working, expedited post COVID but AI just exponentially accelerated asynchronous work. Bulk of the software in 2026 has been written, reviewed and tested by AI (whether the HN crowd likes it or agrees with the practice, or not). All asynchronously.
These two factors have come together where the average tech worker out of SF is not longer very valuable - companies are just hiring equally talented workers out of Austin or Vietnam, for way cheaper, assisted by token spend.
The highly skilled will continue to get hired but the median worker out of San Francisco is being replaced
The doctors and nurses might feel smug and cozy having pulled off a hiest and putting regulation in their favor for now but you know who are even better at pulling off hiests and putting regulation in their favor? MBAs and corporate attorneys.
The fat salaries of doctors and nurses are going to be squeezed out into the hospital's bottomline and the doctors and nurses won't be able to do anything about it because the very drama they pulled off to get to where they are will be used against them. It's already happenning - wages for doctors and nurses aren't growing and they are being eaten away by inflation. It's very likely they will make the same amount of dollars in 2036 that they make today.
Even specialists like surgeons are not going to be spared: https://news.ycombinator.com/item?id=50336138
The tech industry has chosen not to play dirty and we don't have to.
AI is making it increasingly easy to start your own business. I certainly have been encouraging people to find their own customers before their (ex)employer forces them to.
Starting your own business isn't for everyone and it's scary but it's no more difficult or scary than 12 rounds of tech interviews.
Let's continue to make the pie bigger instead of putting restrictions on who can even get near the pie.
but it is to people outside tech. I argue, pretty soon, San Francisco will become a magically expensive place for anyone outside AI
Now a shock to the HN audience: Around 90% of the Bay Area population does not work in tech!
> Today, the rent is going up again
AI is minting billionaires who can afford to purchase a lot of housing. San Francisco is supply constrained on housing.
Pricing of homes in San Francisco are going up because a lot less people are getting really rich and offering millions over asking price for the few homes that are up for sale.
An average person's income didn't go up in San Francisco. They suddenly don't have more money to pay in rent or a mortgage.
The life of an average person in San Francisco is not rosy nor indicative of this growth and that is the point I'm making and evidence that the article provides.
San Francisco is one of the most expensive places on earth. $180k/yr doesn't go very far in San Francisco specially if you want to have a family. Dual income becomes pretty much mandatory and if one of those incomes are lost, things start to get tough.
Pretty much anywhere else in the US (outside some parts of the East coast) and almost anywhere in the world, $180k/yr is an amazing salary.
Tech was already moving towards asynchronous working but AI just exponentially accelerated asynchronous work. Bulk of the software in 2026 has been written, reviewed and tested by AI (whether the HN crowd likes it or agrees with the practice, or not).
"AI" has, and will, suck out capital out of the market - this, coupled with non-zero interest rates is putting companies under pressure where cost savings is the top, if not one of the top priorities.
So these two conversations are happening in every board and executive room in San Francisco:
- SMB/startups: Can we go faster with less people if we load up on tokens? Yes.
- MNCs/Large Orgs: Are our engineers out of Austin or Asia equally productive with AI assistance? Yes.
So the decision is simple: San Francisco's expensive engineers and tech workers are useful in a rapidly contracting niche. They are still valuable for startups based out of SF (but the startups are just hiring less than they used to). They are still valuable for rapidly expanding companies based out of SF that are not yet having the cost discussion - Anthropic and OpenAI are certainly hiring.
But rest of the companies are just hiring equally talented workers out of Austin or Vietnam, for way cheaper.
The highly skilled will continue to get hired but the median worker out of San Francisco is being replaced.
That said, AI is making it increasingly easy to start your own business. I certainly have been encouraging people to find their own customers before their (ex)employer forces them to.
Could it be that Marc Benioff knows a few things that you don't know? If so, any guesses what those could be?
- File taxes and find issues that "accredited", "certified" entities missed. In one instance, my federal refund increased by nearly $5k. At $20/mo, just the refund is two decades worth of value.
- Appeal health, home and car insurance claims successfully
- Find out details of interesting plants without forking out tens of dollars for an app that claims to do the same but fails to do so most of the time
- Summarize never ending email threads with 20 disinterested parties who would rather not be on the email but have to respond to show they are "engaged"
- Write a convincing email to my mother in law why I cannot meet her this weekend and hold her hand through the tough times she's having with her ethnic neighbor who annoys her with their "ethnic" music and "smell of fish" BBQ
- Locate the user manual for my microwave and what that rapid 3 beeps every 1s for 2 minutes that starts at 10AM since the last 2 days mean
You might very well find these useless, trivial or even make your blood boil but my everyday life is vastly better due to "AI"
If you hate marketing, don't do it. If you enjoy software development, have fun doing it.
However, you will remain an employee and you have to appreciate the pros and cons of being an employee. The other thing you have to appreciate is that the number of employees writing code who are AI, are cheaper than any human on the planet and works 24x7 without complaining or asking for a raise is going to go up. Likely exponentially.
This doesn't mean demand for good software developers is going to go down, but increasingly, software development is likely going to look like how one human Amazon warehouse worker wears special harnesses that allow them to lift far more weight, walk longer distances and interact with dozens of robots in the warehouse.
As I said in the parent it's not for everyone but a lot more people who were struggling to start their own venture will now be more capable of and have more freedom.
It might very well turn out that in a few years, the only way to enjoy software development the way you like it, is to establish your own boutique software development company.
An alternative, specially because AI isn't regulated to oblivion yet: just don't use OpenAI. Vote with your wallet. That's it. There's no need to "cap how much processing power these companies can accumulate".
If you are right in your hypothesis, more people will agree with you simply because people in general are smart, and OpenAI will die a natural death.
A lot of software engineers don't write code at their place of employment because they enjoy spinning up pods to serve their Django apps - it gets them paid so they can then do what they actually enjoy.
What AI will allow these software engineers to do, is bypass their employers and serve customer niches directly, capturing more of the revenue themselves.
This isn't for everyone, specially those software engineers who do write code at their place of employment because of the pleasure of it, but its an unlock for those who couldn't just make the leap before.
And that's precisely where AI provides tremendous leverage and why the economy will expand rather than contract.
Find a section of customers your employer isn't particularly excited about serving, find out if they will be your ideal customers and serve them directly.
I can't convince you that AI is good for you just like I can't convince a recently divorced person that love exists or marriage is good or their spouse can be a fantastic person for someone else even if they were a horrible fit for them.
I am really sorry you're unemployed due to AI and the only thing I can really do for you is pray you find a good, stable source of income soon.
Which region is this?
Are hurricanes that reset all the infrastructure a recent phenomena in this region? How did people manage in the 60s, 70s, 80s, 90s etc before these rather large (5+ ton HVAC) were even an option?
If nothing works out domestically, this might very well be where the industry ends up.
The current attempt is to have domestic (in the continental US) AI - because owning the infrastructure is way more valuable than renting it.
The problem with electric and community infrastructure is that very few electric infrastructure is owned by the community, but that's not the core issue because the other infrastructure that is indeed owned by community has people in the community split on what's useful and what not.
Imagine how difficult it is to decide where to go out to lunch in a group of people. Now up the stakes to getting people to agree on whose backyard and where holes and trenches and dug and electric poles and power stations are put up.
Also, just like oil is dirty, bad for the environment and kills people, so will the data centers. And the power plants to power them.
Our options are not to not have Data centers and their power plants but whether to have them built elsewhere (even outside the US) and pay rent or own them and get paid the rent.
Arguments about having more efficient or green Data centers and power plants miss the point. Engineering is a process of iterative improvement - you don't begin with the perfect product, you begin with one that crashes, burns and breaks some bones.
It's understandable though that no one one wants to be "it".
but so does React, Vue and Kubernetes and coding agents are crushing it.
I think if you started your company today, you would have a very different take.
> Because they have the resources to keep it up to date. And even they get it wrong sometimes
are these resources that require intelligence or something else?
Inconsistencies already are known and well studied - this is one of the domains of corporate tax lawyers - how to interpret the inconsistencies into a strategy to benefit the company, so that even if the company is sued by the IRS, a well formed argument would be made and case dismissed or fines significantly reduced.
|----------------|------------------------------:|--------------------:|
| United States | 850,000 | 750,000–950,000 |
| India | 115,000 | 100,000–125,000 |
| United Kingdom | 68,000 | 55,000–75,000 |
| Canada | 40,000 | 30,000–45,000 |
| Germany | 36,000 | 30,000–40,000 |
| Spain | 27,000 | 23,000–30,000 |
| France | 23,000 | 19,000–25,000 |
| Italy | 17,000 | 14,000–19,000 |
| Poland | 17,000 | 14,000–19,000 |
| Japan | 13,000 | 11,000–16,000 |
| Mexico | 13,000 | 10,000–15,000 |
| Australia | 6,000 | 4,500–8,500 |
| Ireland | 6,000 | 5,500–6,500 |
| Singapore | 2,400 | 2,000–2,500 |
That's a lot of "nobody"
Not an employee, just have a "shit ton" of Amazon stock that has been beating inflation very well. The last few quarters of volatility has allowed me to buy and sell, often fully covering my monthly expenses. I'm very greatful for that. Far from a company "nobody" wants to work for. You have a better performing company than Amazon to recommend that I can use instead? I'm all ears and I'll even give you a kickback for your time!
- File taxes and find issues that "accredited", "certified" entities missed. In one instance, my federal refund increased by nearly $5k. At $20/mo, just the refund is two decades worth of value.
- Appeal health, home and car insurance claims successfully
- Find out details of interesting plants without forking out tens of dollars for an app that claims to do the same but fails to do so most of the time
- Summarize never ending email threads with 20 disinterested parties who would rather not be on the email but have to respond to show they are "engaged"
- Write a convincing email to my mother in law why I cannot meet her this weekend and hold her hand through the tough times she's having with her ethnic neighbor who annoys her with their "ethnic" music and "smell of fish" BBQ
- Locate the user manual for my microwave and what that rapid 3 beeps every 1s for 2 minutes that starts at 10AM since the last 2 days mean