Since the time i wrote my last blog post on AI (AI could be Zoom 2.0) around 4 months ago, a lot has changed in the AI world - especially from a markets' standpoint!! Then perceived AI darlings like Oracle, Palantir, Microsoft, Salesforce etc. are down anywhere between 25%-50%! The peak euphoric narrative of "AI will be everything" started seeing some serious reality check. Market participants are now rightfully questioning the (1) sustainability of many "AI native" business models, (2) debt repayment ability, (3) ROI on capex and (4) if they should be bailed out with tax payers' money in case some of them go belly up in the race of unhinged spending on hardware infra with a limited useful life. I don't attribute the entire credit for this market pivot to my blog post though :) :)
Understandably, many of these AI company CEOs and founders seem rattled at such sharp drop in their share prices and the bouquet of high flying IPOs potentially going We Work way. Remember in 2019, We Work too was positioning itself as a fancy 'tech' company demanding exorbitant valuations - while in 'reality' it was just another 'infra' firm. Deja Vu!! So, many of these alpha male CEOs now doubled down on their "AI will replace everything" narrative! "We might be 6-12 months away from AI models doing most or all of software engineers do" said Dario Amodei, Co founder & CEO of Anthropic at WEF-26. If you listen to full interview (link), he tried hedging that sensational statement with some caveats subsequently. But, he very well knows that global media and markets will just pounce only on the headline - which will make a strong case for a likely over hyped IPO at nosebleed valuations. Alex Karp, CEO of Palantir, made another absurd statement that with AI, they are now collapsing SAP implementation timelines from years to weeks!
Ironically, an year earlier in WEF-25, the CEO of Salesforce made a similar statement that his would be the last generation of CEOs working with only humans (link). In Sep-25, Salesforce laid off 4k people because the CEO then needed less heads - thanks to AI (link). Towards Jan-26, multiple news articles started reporting that senior management team of Salesforce now regrets letting go of 4k employees in their quest to force AI agents into work flows (link). This is classic déjà vu of Klarna template (link). Salesforce stock was down almost 45% in last 13 months!! Salesforce now say they are pulling back from their heavy reliance on LLMs for coding after encountering reliability issues that have shaken executive confidence!! The company says they are shifting away from Gen AI towards more predictable and deterministic methods of software engineering :) :) This is after facing serious client backlash over consistency and precision of output generated by it's AI offering - Agentforce.
Seriously?!! It took them 13 months of time, 4k+ lay offs, 45% market cap erosion and serious client backlash to understand the difference between deterministic and probabilstic output of software code? A third year computer science engineering intern would have known this better!! If Salesforce' management is genuinely that naive, it is still pardonable. But, if they did this entire drama to fool the gullible analysts, investors and journalists, they clearly shot themselves in their foot!
Nothing turns on the analysts, investors and journalists more than a eccentric founder saying he is going to "kill the status quo". And they fall for this rubbish every single time with no hindsight learning. It was We Work and Klarna once, Salesforce a year ago and Anthropic / Palantir now. If a software or service that is so deeply intertwined with the arteries and veins of an enterprise can be so easily replaced by a probabilistic language model, my generation of Computer Science Engineers (2012 batch) would have never even heard of something called "IBM main frames." But, here we are in 2026, when IBM's infrastructure segment just reported 17% YoY (CC) revenue growth in Dec-25 - courtesy launch of its latest main frame IBM Z17 a few months earlier! Software and services as business models are lot more nuanced than what these noisy and lousy headlines can often (mis)communicate!
But, who cares about nuance? Klarna got its IPO done selling this "AI will do everything" narrative, only to take a U turn later on. Salesforce tried pulling off the same stunt to possibly pump up its stock - which miserably back fired! Now, its the turn of Palantir, Anthropic, Open AI and SpaceX etc. to give that nudge to their stock prices / upcoming IPOs. Journalists and analysts just need a sensational headline for that day to thrive. Self certified experts, disgruntled ex-employees of tech companies often want to chime in the debate in support of the 'cool' technology which will "kill everything" unless they change. 24 months ago it was "Google is dead", 18 months back it was "BPO / KPO is dead", 11 months back it was "US exceptionalism is dead", 8 months back it was "IT services is dead" and now it is "Software is dead". These narratives and noises come and go!!
But, narratives often change behaviours. This entire panic mongering around lay-offs seem to be seriously impacting the discretionary consumption in India. I continue to believe that AI as a technology could be a great net positive for the economy over the long term - both in value terms and employment. But, if select companies are allowed to set and run a fake narrative, suiting their own interests, there could be negative externalities in terms of consumption behaviour - especially in economies like India which are heavily reliant on tech employment. In that case, the net impact of AI on the economy needs to be studied at greater depth! Even worse, this hyped up AI narrative can eventually become self destructive for the technology when reality sinks in for enterprises - like it happened with Salesforce!
Comments