Tushar Chaturvedi

@unusual_whales All this just to lose out to claude again.

Microsoft, $MSFT, is building a super app that combines coding, chat, and other Copilot AI tools, per FORTUNE.

@WheelieInvestor @_WheeliInvestor Please always keep going. You are an inspiration to all of us!

Why I’m so thankful for the stock market: Ever since I was a young kid, I always worked very hard to make money. I mowed lawns, shoveled driveways, raked leaves. I worked as hard as I possibly could. When I had my skiing accident and became paralyzed, it was a lot harder to make money because so many jobs were inaccessible for me. The Stock Market has allowed me to keep making money and set myself ahead of the crowd. There is no exclusion in the market for people like me. And for that reason, I will be forever thankful for the opportunities I have been given and the chance to keep making money despite the challenges. Final note: I couldn’t be happier with where I’m at and in thankful for everything I’ve worked so hard for

@the_Fitcode Such startups deserve to fail.

Average "Indian mentality"

@UnSubtleDesi This level of hate is extremely disheartening and unacceptable in this day and age.

This is how Jihadis stabbed Surya Chauhan to death. Stabbed him repeatedly till he died. Why? Because Surya, a Hindu, didn’t want to witness an animal being slaughtered during Bakra Eid. These are barbarians who deserve nothing short of a public execution.

As more and more people outsource an increasing proportion of their cognitive abilities, critical reasoning, creativity and imagination to AI, the marginal effort required for obtaining an edge over others by using one's own abilities is only going down.

@talk2anuradha The incident is so gruesome that it's nearly impossible to believe that another human being did this.

> A 15 year old Gopal Sharma > Kidnapped & brutally murdered in Greater Noida > Subjected to unimaginable torture > His eyes and tongue were gouged out > No mainstream media coverage And ZERO outrage from so-called Brahmin leaders in Uttar Pradesh.

@VazeKshitij A lot of people still tend to choose careers in engg that are SWE-oriented rather than manufacturing-oriented. That's just market dynamics though. The real solution is to have more capital allocation towards chip/electronics companies & creating requisite incentives.

By making people understand the CSE (app and web dev) is not the only way of living a comfortable life.

@isaacrrr7 This is extremely sad to see in this day and age.

PAKISTÁN: Cientos de hombres abandonaron su mezquita como zombis para acosar sexualmente a una mujer después de verla caminar cerca de la mezquita sin hijab. Esta es una cultura donde las mujeres no valen nada. Una vez que entiendas eso, entenderás a sus bandas de violadores.

@BoringBiz_ Building a world-class model is a specialist job. I see it as an inevitability for the majority of the law firms to not be able to move beyond the typical fine-tune a specific model for their own use-case ultimately.

This is a massive bull case for on prem vs cloud, especially as more large companies move towards building their own proprietary tools Few realize this yet

@AdityaInvests90 Selling Nvidia and using that to buy AMD, Micron. A little bit or Intel as well. The market already priced in record revenue growth. Maybe its now pricing in things like the large shift to CPU for inference tasks.

Why is $NVDA selling off everyday after earnings even though they reported almost 100% in revenue growth?

@SovMichael It's the new peripheral industries that will be created and inside those peripheral industries there'll new types of jobs and businesses. Similar to when more people could afford cars, it changed business, tourism, travel alongwith public infrastructure, highways, hotels, etc.

In other words, when Altman, Musk, Andreessen and others attempted to sell everyone on a world without work and abundant wealth created by AI controlled robotics, everyone freaked out. Now Altman is going to push another fantasy that there will be additional jobs that are created that apparently will not be affected by the long-planned post-work world these cultists are creating. What you should realize at this point is that none of these utopian wizards should be in charge of your (our) future.

@Polymarket All that AI spending being recovered by the age-old tried and tested SaaS subscription models.

NEW: Meta announces new paid subscription tiers for its apps: Instagram Plus: $3.99/month Facebook Plus: $3.99/month WhatsApp Plus: $2.99/month

@MsMelChen So many countries are going to face trade deficits with a subset of countries simply because the said subset of countries continue to invest and advance in chips, robotics and AI models. Also, a lot of these realizations have come only recently post-LLMs and not prior to them.

Wake up, Europe. You have exactly two choices: America’s tech stack or China’s. You cannot build your own. You cannot complete anything at scale. All you do is pass legislation, issue fines, write stern letters of concern, and hold press conferences about “digital sovereignty” while your cloud infrastructure, AI models, semiconductors, and platforms remain overwhelmingly dependent on foreign technology. Pretending you’ll magically “build European champions” by taxing, fining, and hamstringing the only people who actually ship product at global scale is delusional. Decades of evidence prove it. Brain drain continues. Investment gaps widen. We know what the data says: Europe lags behind in AI, cloud, semiconductors, and consumer platforms because of the regulatory moat you build and maintain around mediocrity. So you can push for sovereignty but it will not materialize because you’re unwilling to do what it takes to get there.

@Lamborghini This is what cars are meant to look like.

Proud to keep you dreaming.

@Wanderlosth LLMs serve to augment someone's capabilities when they know exactly what it is that they are looking for at a working level without having to manually go through large documentation for the actually useful bits. Otherwise, they're intellectual supplements at best.

When you’re doing serious work, it’ll make you realize how useless LLMs are

@KislayParashar1 If optimized correctly, well-written code that uses TPUs via production-grade libraries (such as Jax) can beat GPU code for ML. The market will eventually price this in. However, GPUs are much more general-purpose than TPUs, and the market will price this in as well.

Anthropic just committed to one million Google TPU chips by 2027. This is the same company that built everything on Nvidia hardware. Google cut Gemini serving costs by 78% in 2025 using chips they built internally for ten years. Ironwood, their seventh gen TPU, now matches Blackwell on inference benchmarks. 4,614 TFLOPS per chip. They do not sell it to anyone. Every chip stays inside Google's own walls. Midjourney cut its inference bill from 2.1 million to 700,000 dollars after switching. A Series C startup sold 128 H100s on the secondary market after its monthly costs dropped from 340,000 to 89,000 dollars on TPUs. Nvidia still has 80 percent gross margins, 5 trillion dollar market cap, and CUDA lock-in that took 15 years to build. None of that is gone. But when your best customers start selling your hardware on the secondary market, the moat is not as deep as the market cap suggests. At what point does Nvidia's valuation start pricing that in?

@NewsArenaIndia @siddarthpaim "Using AI" comes down to knowing exactly what you want. Don't really have to pay USD 25,000 to find out what that is.

'AI Gurus' are charging ₹24 lakh ($25,000) per day to teach Wall Street Bankers how to use AI.

@UsmanQazi1 @iam_shimorekato I mean Pakistan.

@Tushar___C @iam_shimorekato By that country, you mean America?

This movie should be widely boycotted.

@valigo Glad that someone said this. Another rudimentary thought experiment would be that if a set of people is vibecoding the majority of their codebase using similar prompts, then doesn't the world end with codebases with similar vulnerabilities (and haven't even checked the code!)

Vibecoding is when you let LLMs generate code and you accept it as long as the end result appears to be working. You don't even review the code, you don't care. When you rigorously ensure that generated code is up to a standard, it's not vibecoding anymore. It's programming.