nick

im bullish on bootstrapping as a teenager at 13 @artfreebrey and i built a minecraft server which made 3k in one month and blew our minds then we moved into sneaker botting then we moved into wholesale then we moved into software each stage teaches you soo much & no need for VC pressure, just have fun printing cash 💰

Holy moly, @particle_api 's model landed on our pareto frontier for data analytics tasks, beating out Terra-6.1 and Haiku 5.5. First time a new player has landed on our frontier in years. Congrats @Suhail and team! Full results here: plotly.com/blog/particle-1-p…

Reading this before applying to @ycombinator will substantially increase your acceptance likelihood. I spent this summer in SF talking to 100+ YC founders, and the pattern for who actually gets in hasn't changed. It's just more obvious now, because this cycle's inbox is flooded with the same idea. 1. The people who get in first try care the least about getting in. Most of the accepted founders I talked to applied in 30 minutes to 2 hours, on a whim. One didn't even realize how hard YC was to get into until he asked Claude and it told him it was a 1% acceptance rate. That's not luck. Partners read hundreds of these. They are trained to detect bullshit, desperation, and unconfidence. The second your application reads like you're begging to get in, you've already lost. If you're spending 4 days wordsmithing, you're signaling the exact thing they're filtering for. 2. Pre-revenue is not a weakness. It's max potential. This was the most counterintuitive thing I heard, and it still trips founders up. Most of the recent batch founders I met got in with just an idea or a prototype. One founder told me he was rejected 14 times when he applied with revenue, and got in the one time he applied without it. Why? YC wants to see a roadmap from zero to $100 million in ARR within five years. That's easy to believe when you haven't sold anything yet. It's a lot harder when you're doing $10k/mo growing 5% month over month. At that point the math already caps you, and it's obvious you won't get to $100M in five years. Don't apologize for being early. Sell the ceiling, not the current number. 3. Niche beats generic, especially right now. A company building autonomous tractors or electrolyte synthesis for spacecraft batteries has 10x better odds than another "agents for X" company. In this cycle that's even more true. If I can swap your company name with 500 other applications and the pitch still works, you don't have a pitch. You have a category. The founders who stand out are solving a specific, painful problem for a specific person who feels it every day. Boring on the surface, obvious once you talk to the customer. 4. It's about the team, not the product. The product will change 10 times. The team won't, and YC knows that. What really sticks out is this combo: you started the idea within the past month, and you already have 1 to 10 paying customers, LOIs, or contracts signed, plus a professional background that shows you've been excellent in the past. That speed is the proof. It shows you can actually execute, not just talk about an idea. Pair that with a team that has done hard things before, and that's what YC is betting on. Put together, a winning application only has to do 3 things, in the fewest words humanly possible: - insight - ability to execute - confidence And behind all three is the team. If a sentence doesn't prove one of those, cut it. If you're applying for W27, don't overthink it. Write it fast, write it straight, and submit it. And if you want a second pair of eyes before Monday, drop a comment or DM. Happy to help.

Meet NVIDIA Nemotron 3.5 Transcribe 🎙️ An open-weight, promptable 30B MoE model with ~4B active parameters that brings transcription in 40 languages, speaker attribution, translation, contextual prompting and audio understanding into one checkpoint. Watch it keep track of who said what when everyone’s talking at once. 1/ 🧵

looking to raise venture capital to buy a 512gb mac studio

Own Your Intelligence @xuster Most companies building with AI still rent their intelligence from frontier labs. As open models improve, more companies will want models they can run, train, and control themselves. The opportunity is to turn a company’s workflows, expertise, and failures into specialized models that improve with the business.

There is an absolute barrage of inference providers popping up that get to $1M/mo extremely quickly. This is the rough economics of inference providers: Even if you rent on-demand B300s at $8/GPU-hr, many open source models (GLM 5.3, Kimi K3, DeepSeek 4.1 Flash / Pro) can get you >$10-20/GPU-hr at reasonable throughputs (100tok/s). DeepInfer from DeepSeek directly shows they can get $38/GPU-hr on off-peak hours. Every chip / model has a different throughput / interactivity curve, but there is a lot of arbitrage between even high rental spot prices and token prices. New chips (Vera Rubins) also change this meaningfully. Your ability to be making gross margin relies on your utilization and how good your stack is. Gateways like OpenRouter, Vercel and more help solve that problem by aggregating demand to you. May inference providers discount their tokens to get longer term commits too. I've even heard stories of selling back Chinese model tokens to Chinese labs! One thorn in the economics of these players is that many Chinese open source model labs are now asking for rev share if you're at a certain scale, espp for pre-release access to their models for day 1 support. That further pushes the threshold to be in the money. The barrier to entry seems low and falling. There are even stories of smaller ones who can "vibe code" a way to stand up OSS models on a cluster they rent. And as a result, we're seeing 10s of such companies pop up almost every week and immediately skyrocketing revenue at low-ish gross margin and raising rounds on the back of that. Unsurprisingly, because the actual service provided seems commodity, this eventually drives up compute prices and drives down token prices. In theory, it would seem like neoclouds who have the lowest cost of compute and themselves have started inference businesses would win the race to the bottom (Crusoe Coreweave Nebius and GMI all have inference businesses), with larger players (Modal Fireworks Baseten) being second because they are well capitalized to make bigger compute commits cheaply and have demand (high utilization), but the reality is the pie is growing so quickly that even upstarts are absolutely crushing it and growing aggressively (inference dot net, DeepInfra, Wafer, Relace, Morph). Strategies vary across the board: neoclouds want to serve inference, inference players want to be neoclouds and do higher margin reserved instances / post training, and some want to go heterogeneous hardware or creatively flexible compute constructs. The risks are: do open source models continue to be competitive? Will their creators rent seek more? Will their shape change compute needs? How do different chips change economics? Who gets compute in a supply crunch? Below is a aggressive sketch of how gross margin scales with utilization for various models / chip sets. For now, there are many inference providers and they’re all ripping!

We're launching a $100,000 challenge to see who can make Muse Spark run fastest: openkernelchallenge.dev/

We're launching a $100,000 challenge to see who can make Muse Spark run fastest: openkernelchallenge.dev/

Het kabinet volgt de situatie rond de kwetsbaarheden in PHP nauwgezet. Het advies van het Nationaal Cyber Security Centrum @ncsc_nl is helder: installeer de patch direct en houd kwetsbare systemen offline tot dat is gebeurd. Websites als DigiD en MijnOverheid komen weer online zodra dat veilig kan.