Category: Thinks

ET: “Ecommerce, quick commerce and food delivery platforms are projected to generate more than Rs 28,000 crore in revenue from advertising in 2026, up 30% from last year, making ads a critical margin lever for these entities working on narrow margins. Ecommerce majors such as Amazon and Flipkart are expected to clock Rs 19,000-20,000 crore in ad revenue in 2026, up from the previous year’s Rs 16,000 crore, according to data shared by Deloitte. Quick commerce platforms, including Blinkit, Instamart and Zepto, are expected to clock Rs 4,900 crore in ad revenue this year, up from Rs 3,000 crore posted the previous year, according to Datum Intelligence. Food delivery giants Zomato and Swiggy are expected to clock 20-25% jump in their combined 2025 ad revenue of Rs 2,500 crore, a senior industry executive said.”

Ashu Garg: “Every enterprise vertical—legal, insurance, healthcare, financial services, procurement, security—has decades of accumulated institutional judgment that has never been structured, never been compounded, never been made operational. That judgment is what makes a $2,000/hour partner worth $2,000/hour. Frontier models are raising the floor, but they’re not raising the ceiling. The ceiling is institutional. It is the accumulated, domain-specific, outcome-tested reasoning about how this organization makes decisions under these constraints. That is what compounds and what cannot be replicated by a better base model. It’s also what’s finally becoming capturable, structurable, and learnable. Consumer giants built trillion-dollar empires by compounding behavioral traces. The enterprise equivalent is just now becoming possible, and the prize is arguably larger. The companies that build the infrastructure to make this real will define the next era of enterprise value.”

TheMaxSource: “Stop measuring satisfaction without measuring usage frequency. Customer Satisfaction Score tells you how people feel after interaction. It doesn’t predict whether they’ll interact again. Track your stickiness ratio instead. Calculate daily active users divided by monthly active users. If this number sits below 20%, satisfied customers are already leaving. Audit your value delivery frequency. If users only derive benefit weekly or monthly, you need either more frequent value moments or mechanisms that create habitual engagement between them. Duolingo didn’t make language learning happen faster. They broke it into daily 5-minute sessions.”

Arnold Kling: “The Moral Dyad model was propounded by Daniel Wegner and Kurt Gray in their book The Mind Club, published in 2016…Their research sought to determine how we view the minds of other human beings. What they found was that there are two clusters of beliefs that we hold about other humans. One cluster concerns agency. We think of other humans as having the ability to make choices, form plans, and work toward goals. The other cluster concerns feelings. We think of other humans as having the capacity to experience sensations. We are especially inclined to notice when other humans feel pain. The Moral Dyad model says that in any moral situation we are inclined to view one human or group of humans as having all of the agency, while the other individual or group feels all of the pain. That is, instead of recognizing that both sides have agency and feelings, we gravitate toward taking an either-or view of the situation. Wegner and Gray use a robot/baby metaphor to describe the Moral Dyad. A robot can carry out intentions but cannot feel pain. A baby can feel pain but is not equipped to undertake deliberate actions. The Moral Dyad model says that we treat one side as if it were a robot and the other side as if it were a baby.”

Nicholas Decker: “Manufacturing firms are much smaller in India and grow much less over time than in America (Hsieh and Klenow, 2014). However, when we look around the world, we see a similar story. People are moving out of the countryside, but into providing services in cities, not manufacturing.”

Derek Thompson: “Phones are global, but what’s on our phones is exquisitely individual. For this reason, overall phone effects are hard to study. They are best understood as a relentless information-delivery system whose utility or harm is exquisitely dependent on the type of information that people access. This might explain why cultures with more anxious or polarizing content—such as the U.S.—see higher and faster rates of anxiety and polarization. Rather than adopt an empirical nihilism about all this (ah, well, phones are complicated, let’s just do nothing!), we should pay close attention to the consistent finding that people tend to be a little happier and little more attentive when they un-hook from the information-IV drip of their personal devices.”

FT: “The problem is that these algorithms are by their nature backward-looking: they serve us content based on what we have already liked. And so we keep on being fed the same diet. We talk a lot about AI slop, but often it’s simply regurgitating the human-generated slop already out there. And so we find ourselves in the grips of what Theodor Adorno might have called a “mimetic regression” — and a ferocious one at that. We must fight back against our algorithmic overlords. If we don’t, we might be stuck in a risk-averse, slop-filled, cultural feedback loop forever.”

WSJ: ““TBPN,” shorthand for Technology Business Programming Network, treats technology news with the seriousness of a sportscaster describing a winning play. It is widely followed by tech enthusiasts, from industry practitioners to AI-curious young people. OpenAI is trying to change long-established habits around how people interact with technology, and fight growing anxiety about the impact that AI will have on the workforce and society writ large. Within Silicon Valley, it is battling for mind-share among young startup founders, software engineers and tech executives whose perceptions are largely shaped by what they see on social media—specifically X. That is where “TBPN” comes in.”

FT: “Social media companies make money from attention, which in practice means rewarding sensationalism and inflammatory content with little regard for truth. They have also until recently avoided liability for harmful or false information by using the defence that they are merely neutral platforms on which other people publish. In contrast, as British philosopher Dan Williams argues, AI companies are competing to serve customers who are paying for accurate, objective and, well, intelligent, tools that deliver factual information, often for business-critical purposes. When LLMs do surface harmful or dangerous content, they are on the hook. In Williams’s parlance, this makes them fundamentally “technocratising”, exerting the opposite force to social media’s radically democratising influence.”

Rama Bijapurkar: “Most plans rest on the logic of “industry is expected to grow at x per cent and, on that basis , we will grow y per cent”. Perhaps boards also need to do their bit in pushing customer centricity by asking (especially if the company is a market leader) how exactly industry growth happens and what customer-related assumptions underpin this forecast. They must also ask for growth plans to be broken down into components of price-led growth, mix (or portfolio)-led growth, and volume-led growth, and then ask “who”— not to be confused with “where” (geography) and “what” (product) — this growth will come from and why, Socratically drilling it all the way down to management’s foresight about customer behaviour. Yes, it works for all kinds of business, be they business-to-customer, business-to-business or direct-to-consumer!”

FT: “For all practical purposes, prediction markets seem to work as well as a gaggle of economic experts. However, sourcing the wisdom of crowds to help with decision-making can also lead to catastrophic failures. James Surowiecki, in his classic book The Wisdom of Crowds, explains that four criteria need to be met for a crowdsourced answer to be wise. Prediction markets are very good at two of them. They help aggregate private judgments into a collective decision and combine inputs from people with local or specialised knowledge. The remaining two criteria are where prediction markets can and do break down. In order for crowds to make more accurate forecasts than experts, there needs to be a diversity of opinion, and the opinions of the people in the crowd need to be independent of each other. These conditions are typically not violated when crowds try to predict technical issues like inflation or unemployment numbers.”

Peter Earle: “History suggests that the economic consequences of sweeping technological change hinge less on the invention than on the institutional ecosystem surrounding it. Electrification required factory redesign. The internal combustion engine required road networks and suburban development. The Internet required specialized software, new legal frameworks, and payment systems. Artificial intelligence will be no different. Its aggregate productivity impact will depend on education systems that adapt, firms that reorganize workflows, and regulatory regimes that neither stifle experimentation nor generate moral hazard. In that sense, the Productivity Panic of 2026 is likely to be less about machines replacing workers than about whether our institutions can evolve as quickly as our technologies.”

FT: “Meta, Google parent Alphabet, Microsoft, Amazon and Oracle [are] forecast to deploy $4tn of capital expenditure over five years, according to analyst estimates gathered by Visible Alpha, most of it on data centres they hope will reshape their businesses.”

David Oks: “When all is said and done, and the final accounting is made of all human ambitions and achievements and follies, and the final historian turns to that strange realm of human endeavor that we call “computing,” that strange enterprise that gradually grew to encompass an unbelievable share of human life and redefine the entire world around its logic: what will that final historian have to say? Probably they will start with the forerunners, with Llull and Babbage and Lovelace; and then turn to the true pioneers, to Turing and Church and Shannon and von Neumann; and then the masters of hardware, Noyce and Kilby, and of software too, Ritchie and Dijkstra; and eventually they will arrive at PageRank, recommendation systems, neural nets, the transformer architecture, and whichever system ended up bootstrapping itself into superintelligence and thus inaugurating an entirely new epoch of history. But somewhere in their chronicle of this grand arc, for at least a few pages, they will have to talk about the electronic spreadsheet.”

ICONIQ GTM Report. “Companies continue to diversify their GTM strategies by blending top-down and bottom-up motions, with high-growth companies increasingly leaning into bottom-up as a growth lever alongside upmarket expansion. As a result, revenue mix is becoming more balanced across motions; while direct sales and channel remain dominant, high growth companies expect to gain greater leverage from self-serve relative to peers (~20% vs. ~10% respectively).”

NYTimes on Silicon Valley’s ‘Tiny Team’ moment: “As artificial intelligence takes on more and more tasks, tech executives are embracing teams as small as two: one person plus A.I.”

Ben Thompson: “The truth is that Apple’s lack of investment in AI was always going to be a short to medium-term win: the company doesn’t have to spend on infrastructure, and everyone still needs a device. The real threat is in the long-term: what happens if AI becomes so good that it obviates traditional user interfaces? Or, to put it another way, what if the point of integration that is most compelling is not a traditional operating system and hardware device, but rather AI and a dedicated device?”

Ethan Mollick: “AI capability has been running ahead of AI accessibility. The models have been smart enough to do extraordinary things for a while now, but we’ve been making people access that intelligence through chatbots. And, as that cognitive load research shows, the chatbot format is actively working against them. As interfaces improve, we’re going to see what happens when a much larger number of people can actually use what AI is capable of. Every new interface that closes even part of that gap will feel like a leap in AI capability, even when the models haven’t changed (though they are still changing). My guess is that a lot of the “AI disappointment” people sometimes express comes not from the AI being bad, but from the interfaces being wrong. We built one of the most powerful technologies in recent history and then made people access it by typing into a chat window. That will change soon.” [via Jaimit]

WSJ: ““The Laws of Thought” is…a rigorous and captivating account of how cognition can be modeled via three mathematical frameworks: logic, artificial neural networks (“mathematical systems that emulate the operation of the brain”) and probability theory. It’s a “quest for a mathematical theory of the mind,” as the subtitle puts it—opening with Aristotle and ending with artificial intelligence.”

Logan Bartlett: “Across the internet, cloud, and mobile eras, the companies that became durable winners were predominantly founded in years 4 and 5 of each platform shift. Google and Salesforce after Netscape. Snowflake and Datadog after AWS launch. Robinhood and Coinbase after the App Store. ChatGPT launched in November 2022. We are in year 4. It certainly feels possible that OpenAI and Anthropic capture the lion’s share of value this go round, but the last 3 transitions were largely consistent in this. This is also the most crowded, fastest-moving, highest-bar environment I’ve seen, with rounds closing in days and valuations that require exceptional execution to justify from day one. It’s hard not to be excited and uncertain every day.”

Andy Kessler: “Even with discipline, the reason “no side quests” is such a bad idea is that success comes via surprises. Progress demands surprises. Even as of a year ago, OpenAI’s ChatGPT was the undisputed leader in AI. But Anthropic, a company premised on model safety, wasn’t sleeping. As a side project, engineer Boris Cherny built a prototype to control Spotify with Anthropic’s AI Claude. It spread rapidly inside Anthropic and updated to read and write local files and write code. Claude Code tracks its coding mistakes and, in effect, teaches itself how to write better code automatically from simple prompts. Anthropic went from a $4 billion run-rate company nine months ago to $19 billion today. OpenAI is scrambling to catch up. Surprises happen again and again. Penicillin. X-rays. Post-it Notes. Arno Penzias at Bell Labs won the Nobel Prize in Physics for discovering “cosmic microwave background radiation” from the Big Bang.”

TheMaxSource: “Companies rarely fail because the idea was bad. They fail because the internal pieces that must work together never align…Think of [the McKinsey 7S model] as a management X-ray. It shows what is working, what is contradicting and what is missing. When these elements reinforce each other, organizations gain clarity, speed and competitive fitness. When they are misaligned, growth stalls, execution breaks down and leadership has to constantly push instead of the business pulling itself forward.”

TheGreySwan: “The Inversion Point [is] that precise moment when a system’s deepest strength—its most optimized feature—mutates into its most binding constraint. It is the point when/where our past successes become the walls of our future prison.”

Shankar Sharma: “True investment genius isn’t so much about getting bull markets right. Most do. Oraclehood is conferred only when you get bear markets right. That’s Buffett’s cloaked edge that nobody told you about. And anybody with that wizardry won’t be wasting his time managing your money for 1 per cent per year.”

NYTimes: “With seeds, supplements and gadgets (but little expert guidance), Americans of all stripes are seeking wellness through what they eat.”

FT: “[Indian] cinema attendance last year fell 6 per cent from 2024 levels to 832mn, the lowest in a decade barring the pandemic years, according to Ormax Media, a consulting firm tracking India’s entertainment sector…“We are combating gaming, we’re combating streaming, we’re combating sport, we’re combating attention span, because of reels and Instagram,” says Karan Johar.”

Arnold Kling: “Here is one of the puzzles: trading volume in the stock market. Rational investors would not trade often. If you believe the Efficient Market Hypothesis, then you do not assume that you can outsmart the market. Trading to try to beat the market is a losing proposition. In fact, the overwhelming majority of individual investors do not trade frequently. But trading volume far exceeds anything that a model would predict, assuming rational individuals. A small minority of individuals account for most of the individual trading. But institutional investors appear to account for more trading than do individuals (reliable data are not easy to come by). One economic model is that there are “noise traders,” who make trades on the basis of useless or misleading information. Opposite them are market sharks, who buy what the noise traders sell and sell what the noise traders buy…We cannot think in terms of the market acting like a single man, rational or otherwise. Market outcomes reflect the bets made by irrational men with many different beliefs and strategies. The net result is that stock prices cannot be systematically predicted. So the market as a whole is weakly efficient, or weakly rational if you will.”

Bloomberg: “[Xiankun] Wu, 31, designed Junior for almost any business, equipping it with the ability to tap into company data and communications challenges, along with the organizational memory it needs to know who does what and how colleagues are connected to each other. Wu is now courting global corporate customers, offering Junior as a full-fledged AI colleague capable of managing work processes within small and medium enterprises — at a cost of $2,000 a month. Junior has its own phone number, email and Slack account. It can join every Zoom call. “Getting used to the AI agent can be exhausting,” said Wu, who splits his time between Silicon Valley, Hong Kong and Shenzhen…The proposition is blunt: it’s labor, but AI-defined. Junior drafts marketing campaigns, updates customer relationship management systems, monitors inboxes, tracks deadlines across departments and generates reports. It does so proactively: Instead of waiting for prompts, it scans internal communications, identifies gaps and relentlessly nudges employees to close them”.

FT: “If the lessons of previous periods of wrenching change are anything to go by, much of today’s business establishment will muddle through, even as a handful of new AI giants rises to become the most visible winners from the technology. Economic history has shown that “most industrial change happens by very large new companies doing big things,” says Bradley. “And then the rest of the economy kind of just carries on.” Behind that “just carrying on”, there will be no shortage of upheaval, churn and disruption.”

WSJ: “In the provocative, ambitious and exasperating pages of “The Story of Stories,” Mr. Ashton, a former researcher at the Massachusetts Institute of Technology, makes the case that stories have been part of human history so long that they predate language itself. A story, he writes, is a “primary vehicle for emotion” with three components: character, chronology and consequence. Into this form, we put our relationships, our inner lives and what we believe to be our external reality. We live and breathe stories: We concoct them; we relate them; we react to them. From time to time in history, there has come a technological leap that allows greater dispersal of stories: the invention of the printing press; the development of wood-fiber paper; the innovations of radio, television and, of course, the smartphone. The author refers to the manner of these jumps as “saltation,” a geological term pertaining to the movement of particles by wind or water that he likes so much he uses it 11 times.”

FT: “Complex animals appeared on Earth even longer ago than thought, according to a more than 500mn-year-old fossil trove uncovered in south-west China that shows the power of symmetry in evolution. The discovery dates the surge in oceanic evolutionary diversity known as the Cambrian explosion to before the start of the eponymous geological period 539mn years ago, says research published [recently]. The find captures a critical moment when simpler creatures began to evolve into more advanced ones with capabilities that would eventually enable them to thrive on land.”

From “The Infinity Machine: Demis Hassabis, DeepMind, and the Quest for Superintelligence,” by Sebastian Mallaby: “At the end of January 2014, Google bought DeepMind for $650 million, a bargain by today’s standards. But the real payoff for Hassabis came over the next decade, as Google poured billions into DeepMind’s research. The quest for superintelligence, which Hassabis had harbored since his teenage years, would soon go into overdrive.” [WSJ excerpt]

Jacob Mchangama: “Democracies have always worried about dangerous ideas corrupting the young. Intellectuals and lawmakers should absolutely be concerned about how and when our children navigate social media. But they should also be concerned about whether, in our rush to protect our children, we are building an infrastructure of surveillance and censorship that will ultimately threaten the hard-won freedoms we want future generations to enjoy.”

Mike Munger: “Liberalism has two mutually reinforcing aspects. The first is humility: I can’t assume I’m right. The second is toleration: I can’t assume you’re wrong.” [via Arnold Kling]

Ashu Garg: “At moments like this, the biggest risk for founders is not acting quickly enough. Action is what creates opportunities for learning, and your pace of learning is one of the most important variables you control. This also means you need to stay low ego, and be ready (or even actively plan to) to throw out aspects of what you’ve built as the technology improves.”

WSJ: “AI-assisted stories accounted for nearly 20% of Fortune’s web traffic in the second half of 2025. Most were written by [Nick] Lichtenberg.”

FT: ““The economy is rewarding scale like never before,” [Larry] Fink noted, observing how the leading companies in many industries were surging ahead while the rest struggled to keep pace. That trend is anatomised in a new report from the McKinsey Global Institute on market competition. The institute had previously identified the strengths of “wizard” companies that could use their technological magic to outperform “muggle” competitors, in the language of Harry Potter. Chris Bradley, one of the report’s authors, suggests nine “super-wizard” companies are now emerging that command enormous resources and are set to dominate many of the 18 fastest-growing markets of the future, such as ecommerce, AI software and services, space, robotics and autonomous vehicles.  These “omniscalers” (to revert to more traditional McKinsey-ese) include six US companies — Alphabet, Amazon, Apple, Microsoft, Meta and the Tesla/SpaceX cluster — and three Asian giants — Alibaba, Huawei and Samsung. They are all characterised by massive spending on research and development and a proven ability to spin their technology, data and infrastructure expertise into new markets.”

Srikanth Nadhamuni: “Her name is Lakshmi. She lives in a small village in the Krishna delta, Andhra Pradesh, where the fields flood in September and the nearest bank branch is a one-hour bus ride away. She wants to buy a buffalo, not as an aspiration, but as a business plan. A government scheme could fund it. But she cannot read or fill a form and has no one to help her navigate the paperwork. The bank exists. The scheme exists. The money exists. The gap is not financial. It’s procedural. And procedural complexity, invisible to those of us who navigate it daily, is one of the most efficient destroyers of wealth in India. This is the problem at the heart of a white paper I co-authored with colleagues from MIT, IIT Kanpur, IISc and other institutions for the India AI Impact Summit, titled ‘Doot: The AI Agent for Every Indian Citizen.’ It outlines the architecture for an AI agent designed to work for every Indian citizen.”

Reid Hoffman: “What is genuinely true (and exciting) is that software must now incorporate AI generativity as a core feature of its value proposition. The new competitive moat isn’t built from how well a software system’s AI is tuned to the specific needs of its category. A CRM company that ships a deeply intelligent set of agents that iteratively refine your sales workflow, that understands your pipeline more comprehensively than any human analyst, that comes with powerful backend libraries purpose-built for that domain has an extremely well-crafted moat. The incumbents who understand this will evolve. The ones who don’t will be the ones who actually die. But even they will die more slowly than most assume.”

NYTimes: “A long time ago, in England as well as America, people understood a constitution to be like a garment, tailored to fit the body of a nation and intended to “align the character of the land and people it governs with an appropriate frame of government.” This old understanding was universal among the framers, whatever else they disagreed about. So, too, [Mark] Peterson reminds us, was the belief that when a constitutional relationship goes awry — when the garment no longer fits the body — the people have the power, right and responsibility to alter it. Whether we possess the political will to create a new constitutional order better suited to address the challenges of our time seems entirely less certain.

WSJ: “Since the 1970s, engineers speculated this might allow humans to store vast quantities of energy more or less indefinitely. Two problems: At the time, renewable energy cost too much to make it affordable, and adding water usually turns quicklime into an unwieldy goop. A 10-person startup called Cache Energy, working out of a 10,000-square-foot facility in Champaign, Ill., says it has figured out how to make such a cement battery durable, efficient and affordable. The company’s approach is to form cement into tiny balls, each about the size of a kernel of corn. Its engineers add a binding agent—secret though widely available, they say—to keep the balls in shape during the discharge and recharge process. Recharging the pellets requires heat, generated from electricity. When it’s time to discharge that stored energy, adding the right amount of water causes the pellets to release enough heat to generate temperatures up to 1,000 degrees Fahrenheit, says Cache’s founder, Arpit Dwivedi.”

Julia Angwin: “Compensating people for the harm caused by their products is just the silver lining. The real win would be if the social media giants were finally forced to design less harmful products. I’m talking about features like infinite scroll, which entices people with seemingly endless content, and autoplay, which automatically starts videos before our eyes. And of course, there are the algorithms that spread misinformation and amplify outrage. These are all techniques Big Tech uses to keep us staring at the screen for as long as possible. Too bad if its profitable practices extract a terrible cost on its users and on our society.”