Data can be tortured to say whatever you want it to say. Hence, many people nowadays believe AI is leading to job loss while many say AI is not.
Logic, however, seems clear: AI is not good for jobs and career prospects of an average office worker.Image credit: ReutersAs we have seen it happening in the last couple of years, every few months we get to hear about more job cuts. AI is always at the centre of it.
This week again jobs and AI were the topics of discussion in the tech world. Thanks to Microsoft.
The company announced an extraordinary voluntary exit offer, targeting nearly 60,000 employees across departments. Software engineers received emails asking them to consider voluntary retirement.
And so did employees in departments like HR and Legal. Nearly 27 per cent of the entire Microsoft workforce has been put on a — soft — notice.
This is just a voluntary retirement scheme. But it is also Microsoft telling over one fourth of its employees they are no longer needed.
Probably because AI can do their job.Just a couple of days later, CEO of Proton, a company that makes privacy-focussed tools like VPNs and secure email service, said 2026 would be the last year he would hire an engineer. Apparently AI — largely Anthropic Claude — is generating almost 90 per cent software code within Proton.
They don’t need software engineers anymore. Instead the plan, as reported by 404 Media, is to give AI tools to product managers and let them handle entire work from A to Z.The overall crux of the matter seems to be that AI agents, driven by tools like ChatGPT and Claude, are now so good that many office roles and work can be done autonomously.
The companies still need people, but fewer, and mostly to manage AI agents. Sridhar Vembu, who nowadays misses no opportunity to weigh in on the talking point of the day, agrees.
On Thursday he wrote on X: “Any job performed on the computer has this risk now: one smart person using AI can perform the work of 10 people. The risk is getting ever more real.
The only thing we can do is to accept that it is coming and be prepared for it.”So, is AI taking the jobs of people? Different people will answer differently.
Data is all over the place, giving different conclusions depending on how it has been massaged and cajoled. There are areas where job loss is apparent, such as customer care.
Can you recall when you last spoke to a human on a support line at apps like Amazon, Swiggy and Zomato? Then there are areas, such as cyber security, where companies might be hiring more.
If you look at the departments that have been impacted by Microsoft’s voluntary exit offers, you will notice cybersecurity teams have not been touched. But to see whether AI is impacting jobs or not, ignore data for now.
Look at the job market and talk to people who are trying to find a job. It is a very tough scene out there.
Even if the number of explicit layoffs is down relative to how it was a year ago, jobs aren’t available. Companies are quietly letting people go and are no longer hiring.
A lot of companies are no longer adding new jobs. A lot of companies have replaced freelancers and outside consultants with AI agents, particularly in sectors such as graphics designing, communications and writing.
These people, who were anyway on the margins of the formal job sector, are no longer finding work. AI is changing knowledge work at a fundamental level.
Consequently, it is also changing the social contract we humans have with society as far as work is concerned. What sort of future will this change bring?
I am not sure. Nobody is.
Although Francis End-Of-History Fukuyama recently wrote AI will replace human labour in “white collar work that involves symbolic manipulation” that is to say “people sitting behind computer screens all day.” Essentially, one thing is certain: people will have fewer opportunities to do 9-5 jobs in an office. Will it be a good thing or bad?
That depends on how our governments, and we the people, navigate the new realities. Google, AI, Reddit and death of open internetJust as AI threatens jobs, it is also threatening the internet as we have lived it.
There are several forces at work. This week we got a glimpse of one of these forces: Google and AI summaries.
Google, which is the beating heart of the internet, is pushing more and more AI summaries in its search results. These summaries, as you must have noticed, sit on top of the Google search results and give users information they are trying to get right on the page.
Yes, sources are cited in the AI search summaries. But most of the time, there are no clicks.
This is beginning to kill the internet. Search traffic — which is essentially almost all the traffic for any website that is not Top 100 in the world — is down in a big way.
Reduced search traffic means reduced revenue from Google ads. From e-commerce sites to media sites, everyone has been impacted.
It is even impacting Google. The company reported its quarterly results earlier this week.
Its earnings fell short of expectations, and the company acknowledged AI summaries would upset the cart, at least in the short term. Its stock tanked within hours, wiping off over $500 billion from its market cap.
But Google has no other option but to push ahead with AI summaries because otherwise it runs the risk of losing users to ChatGPT and Claude. Also, Google is a tech giant with fingers in many pies.
It will be most likely alright. The websites that depend on it for their business might not be.This brings us to Reddit.
Google earlier used to send people to respective websites. Now, it is summarising information on its own.
But Google is not alone. All AI tools, including ChatGPT and Claude, are doing the same.
They are also using all the data that is generated on a daily basis on the internet to train their AI. In almost all cases — exceptions exist — they don’t pay anyone for this data.
Reddit this week announced that this scooping of its data stops now. In the last few months Reddit has tightened access to its data gradually.
The company now forces most people on mobile to download its app, instead of serving information on a web page. It has killed the OLD url that allowed people a somewhat neater access to information.
This week the company announced it would be disabling all access to Reddit unless you are a registered user. This is extremely hostile to users, but Reddit says it has no option because of AI.
We can argue the merits of the move Reddit is making and how it is contrary to everything that the old internet — largely open and free-flowing — promised people. But AI is forcing everyone to rethink how they meet and serve their users.
And it is forcing everyone to tighten access, put doors and locks. Even Wikipedia is exhorting people to download its app.
Forced logins are now common across the internet. Paywalls are going up at an increased pace.
Whether for security reasons, or business reasons, access to a huge chunk of the web is no longer free and unhindered. In a way the internet of 2026 is very different from the internet of 2016.
And it is, I have a feeling, going to be a very different place by the end of the next year. Anthropic eyes $2 trillion valuation and may get itBoth Anthropic and OpenAI are eventually going to go public.
Anthropic is the first one out of the gate. The company this week filed S-1 with the US government ahead of its IPO, which is now expected around November 15.
The document reveals a few interesting aspects of Anthropic, as well as gives us a glimpse of the AI world. Let me quickly highlight a few details I find interesting: 1- Anthropic is doing surprisingly well in terms of its business.
The company has lost little over $80 billion in the first six months of this year. The number seems huge and devastating, until you look closer at it.
Most of this loss is, according to analysts, paper loss. It is something to do with the re-valuation of some earlier commitments from Anthropic.
The more significant part, the actual cash loss, is less than $ 1 billion. In fact, Anthropic has made nearly $1 billion profit in the last quarter and its quarterly revenue is now over $16 billion, compared to a little over $1 billion in the same quarter in 2025.
In other words, solid growth for a company that is barely five years old. 2- Anthropic holds lots of cash, a little over $123 billion.
It actually has more cash than companies like Google, Meta, Microsoft. 3- Anthropic, after its IPO, is still going to be very tightly controlled by Dario Amodei and his sister Daniela Amodei.
The tech founders are going for a holdings pattern — with multiple classes of stocks — that will allow them to keep total control on the company, similar to what Mark Zuckerberg has done over at Meta.My impression is that this will be the same for all big companies going for IPO in future, including OpenAI. These will be extremely tightly controlled companies where founders will create mechanisms to ensure they can’t be ousted in any circumstance.
4- Anthropic IPO is going to make tens of billionaires and thousands of millionaires because of stock grants its employees have received. This is a possibility even if the valuation doesn’t hit $2 trillion that Anthropic is aiming for.
5- Anthropic is making a lot of money but it is also paying a lot of money — now and in future — to companies like SpaceX, Google and Amazon. That is because it is using their data centres to power its Claude.
If Anthropic thrives, Elon Musk, Sundar Pichai and Jessy Andy would be smugly happy men. Sarvam changes track?Sarvam, arguably India’s best AI company, is making some moves.
But sadly it is not — at least yet — the time for a frontier AI model from the company. This week we heard that Sarvam has entered into a partnership with the Indian government to build AI systems and solutions for the Indian Army. “Sarvam exists to build AI in India, for India,” said Pratyush Kumar, Co-founder, Sarvam. “The core of that mission is to put the country’s own intelligence into the systems it depends on… our team is committed to delivering for all three (Navy, Army and Air Force) Services.”This is a good move and comes at a time when there is chatter of Cognizant thinking of investing in Sarvam at a valuation of nearly $4 billion.
But it also started another chatter — that Sarvam is giving up on its core mission of building a frontier AI model, something in the class of ChatGPT, Claude or even DeepSeek, in India. I believe both objectives, creating solutions for the Indian military and building a frontier model with 2 trillion parameters, can coexist.
But I am also mindful of the fact that building a frontier AI model in India, because of many reasons, is extremely difficult. The dice here, unlike in China or the US, is loaded against anyone trying for it.
It is possible that in its quest Sarvam has run into insurmountable challenges. And that going after lucrative government contracts now seems a wiser and more pragmatic path to business and survival.
Sam Altman goes after RAMOpenAI CEO Sam Altman loves to compare AI with humans. Last time it was on the matter of resources used by AI and human babies!
Now, he is eyeing RAM. A few days ago, he said: "It's quite terrible to have everybody have a bunch of the world's limited supply of DRAM sitting in their own computer and being used 5 per cent of the time… You want maximum utilization in the cloud if you're trying to deal with this very supply-constrained environment.
And somehow many people's intuition is the opposite of that.”In a logical sense, the way it is applied in Silicon Valley, Altman is right. Resources need to be optimised, at every scale, from individual to global.
Valley people are crazy about optimisation. But this is the precise problem with Silicon Valley.
Humans are not logical, we have other needs as well. We can be irrational, like buying a 64GB Mac Studio, just to use Reddit and Discord on it.
Valley bros don’t understand it. They want to optimise us, our lives, the entire world.
And then they wonder why so many people hate them.SIX STORIES YOU CAN'T MISS01SaaSpocalypse may be over, TCS and Infosys prove it with AI revenue02Anthropic IPO: Meet 4 Indian-origin leaders handling its finances, tech and AI governance03Sundar Pichai launches Gemini 4 Carbon, says not everyone will get it04From Vizag to Apple, this IIT Kanpur girl is now using AI to help people deal with chronic illness05Cisco gives Gen Z employee unlimited AI tokens to rewrite all of company's code06Neil Shah owns no Anthropic stock, but he can decide who runs the AI lab