Kishan Joshi

@anirudhamudan @atherenergy Imagine in future if microsoft develops such ev vehicles lmao

Hey @atherenergy, whenever there's a software update, can you send a notification to my phone in the morning so I know I have to turn my scooter on early and wait for it to complete?

@anishmoonka Open Ai is so open now

OpenAI just raised $110 billion at a $730 billion valuation. That’s 56 times its 2025 revenue. For comparison, Apple trades at roughly 9x revenue. Google trades at about 7x. But the interesting part is where the money actually goes. Amazon is putting in $50 billion. OpenAI already owes Amazon $38 billion for cloud servers. NVIDIA is putting in $30 billion. OpenAI is buying NVIDIA’s chips for 10 gigawatts of data centers. SoftBank is putting in $30 billion. SoftBank is also building the $500 billion Stargate data center project OpenAI will run on. The money goes in one door and out the other. It’s called circular financing. Amazon hands OpenAI cash, OpenAI hands it right back for AWS. NVIDIA hands OpenAI cash, OpenAI hands it right back for GPUs. Tech analyst Charles Fitzgerald told Fortune the structure is “a framework, not a deal.” The valuation math is wild on its own. OpenAI was worth $29 billion in January 2023. Three years and five funding rounds later, it’s worth 25x that. It jumped from $157 billion (October 2024) to $300 billion (March 2025) to $500 billion (October 2025) to $730 billion now. Each jump funded by the same companies that happen to be OpenAI’s biggest suppliers. And the bills keep coming. OpenAI originally told investors it had $1.4 trillion in infrastructure commitments over eight years. Last week, it quietly walked that back to $600 billion by 2030. The company burned $8 billion in cash last year and is projected to burn $17 billion this year. HSBC estimates it still faces a $207 billion funding gap even after this round. The growth underneath all this is real. 900 million weekly active users, 50 million paying subscribers, revenue that jumped from $3.7 billion in 2024 to $13.1 billion in 2025. But only about 5% of users pay. And OpenAI owes Microsoft 20% of all revenue through 2032. This is the largest private funding round ever, nearly 3x OpenAI’s own record $40 billion raise from last March. Every dollar raised comes with a spending obligation to the company writing the check.

@diabrowser Interesting, A company abandons a really great project, starts another one, gets funding, and reproduces the same thing in a different repository or folder. Either, as hardcore Arc lovers, we are not able to see the whole picture, or it’s a disastrous decision.

Out of sight → out of mind → stalled PRs. Live Tab Groups are here to break that habit.

@wbetiago The probability of them not finding what they’re looking for is higher in such cases. I’d try to understand what they’re missing and close the cycle with them.

What would you do with such a message form a potential big B2B client?

@dannypostma That's real business 👏🏻

@Pirat_Nation Does that mean it was always possible but it was never done ? 🤔

CEO Jensen Huang says Nvidia could potentially resurrect old GPUs to address shortages and high pricing. Adding performance-boosting advanced AI features to older architectures is also on the table. "Yeah, possibly. And depending on the generation, we could also bring the latest AI technology to previous-generation GPUs. That would require a fair amount of engineering, but it's within the realm of possibility. I'll go back and take a look at this. It's a good idea."

@dannypostma You're absolutely right. Once you begin automating people's complaints, it can trigger a cascading effect that erodes customer trust.

@ycombinator Alex is a really good ai recruiter. The way it interacts with humans feels like you are talking to an actual human. Congratulations team 🎉

Congrats to Alex on their $20M Series A! Alex has become the go-to AI recruiting partner for recruiting teams. With Alex, employers can interview everyone, filter out the unqualified and fraudulent candidates, and shortlist the best. techcrunch.com/2025/09/29/ai-…

On weekends, people are more in that research mode, not just ticking off tasks. Traditional search engines are quick but not great for deep research, comparing info, or aggregating information. GPT makes it way easier to explore topics, compare options, and pull everything together.

@DanBochman Maybe it’s about distribution - Google is just one tap away. It’s the default search engine on browsers, on Android phones, or maybe it’s just habit

sotapapers.com/papers Sota papers is really good alternative of paperswithcode

@justbuilding @bangkokbuild When we rush to launch proven products, it’s mostly about marketing and distribution—not tech. If you’re good at both, go for it. Otherwise, I’d just find a niche market and try to grow it.

@bangkokbuild Ive built out an image editor/generator just didnt release it. It was a chore and I had the claude stoppage. Then I got done and didn't see the great angle I wanted. So its on the shelf for now.

@kylegalbraith What are they trying to sell ?

LinkedIn notification: "Gartner Account Executive wants to connect" Me: *declines* *5 more Gartner AEs immediately appear* It's like a hydra but for enterprise software sales. Cut off one head, two more pitch you their Magic Quadrant.

Spent 2 days debugging why ViTs fail on Oxford-IIIT but shine on MNIST. Turns out: transformers are super data-hungry. After adding DeiT-style training tricks, accuracy jumped from 20% → 45% (better, but still far from MNIST’s easy 95%). #Transformer #vision

@DanBochman @SDinesh91 @fashn_ai @ayaboch Congratulations @SDinesh91 🎉 its really great company, combination of really good people and product

Excited to share that I've joined @fashn_ai as a Founding Engineer! 🚀 I am thrilled to help build the best virtual try-on technology & work on challenging technical problems. Big thanks to @ayaboch and @DanBochman for the opportunity 🙌 I can finally share technical details on what it takes to build a robust system that is reliable and scalable. Keep an eye out for my posts on Twitter/X and FASHN AI blog. This wouldn't have been possible without @DmytroKrasun and @thelifeofrishi 🙏

@ericciarla @firecrawl Already did , pls check

@robokishan @firecrawl Very cool! Dm!

Want to be an intern at @firecrawl? DM your GitHub!

@__Tkat__ Maybe building this on top of brave will be cherry on top

I built a browser

@geohotarchive crazy, add some sci fi thing like upgrade or recursion

chapter one geohot.github.io//blog/jekyll/u…

@___meetpatel___ The sweet spot lies between innovative ideas and market needs. Finding and filling that gap is the key to a successful venture

After building 10+ different ideas (mostly personal - never published), I realized that most ideas are trash. The reason is that what you think people want is vastly different from what people actually pay for. And optimizing for pay is a disaster for long-run business!