~nextbigfuture-com | Bookmarks (363)

What if Tesla succeeds with robotaxi in June 2025 in Austin? Will there be a buying frenzy for Tesla Robotaxi?

Tesla has released a video showing robotic cleaning of the Tesla Cybercab. This will greatly improve the economics of robotaxi. Farzad and others have estimated cleaning as a potentially big cost of Tesla Robotaxi operations. If cleaning with robots drops the cost of robotaxi operations to $100 per month for 3400 miles of operation then ... Read more

Some believe DeepSeek is so efficient that we don’t need more compute and everything has now massive overcapacity because of the model changes. Jevons Paradox is closer to reality because demand has already increased H100 and H200 pricing. Deepseek and High Flyer have a mix of 50,000 H20s, H800s, A100s and H100 GPUs. Deepseek has ... Read more

META CEO Mark Zuckerberg predicts that 2025 is the year that an AI assistant will serve over one billion users and he thinks META will be company that provides that AI assistant. In 2023, Meta revealed they have AI inference accelerators that they are designing in-house specifically for Meta’s AI workloads. Deep learning recommendation models ... Read more

Tesla has announced they they will have paid robotaxi without human drivers in Austin in June 2025 and they will thousands and perhaps ten thousand Optimus teslabots this year. Third megapack factory is already being built but they did not say where yet. Tesla will also deploy robotaxi to California and other locations in the ... Read more

Tesla has its 2024 update on financials and operations. Plans for new vehicles, including more affordable models, remain on track for start of production in the first half of 2025. These vehicles will utilize aspects of the next generation platform as well as aspects of our current platforms and will be produced on the same ... Read more

NuScale and Standard Power are developing two new nuclear reactors facilities in Ohio and Pennsylvania powered by its SMR technology. The plan is to build 24 units of its 77 MWe modules, producing 1,848 MWe energy from both sites. Standard Power thinks this facility could be operational by 2029 at the earliest. Nuscale has developed ... Read more

SpaceX Starlink and Apple have partnered to bring Starlink direct to cellphone connections to iPhone. Starlink works right from your pocket. T-Mobile to beta testers: “You can now stay connected with texting via satellite from virtually anywhere” While it’s text-only for now, SpaceX and T-Mobile plan to add voice calls and data in the future. ... Read more

Tesla cars will now have their first drives from the Factory to the loading dock without human drivers. Teslas now drive themselves from their birthplace at the factory to their designated loading dock lanes without human intervention One step closer to large-scale unsupervised FSD pic.twitter.com/Aj6dHsLaRO — Tesla AI (@Tesla_AI) January 29, 2025 Barry Glack says ... Read more

Jim Fan tells AI community don’t worry. Be happy…and grind on code 24-7. Faster is the only way. Many tech folks are panicking about how much DeepSeek is able to show with so little compute budget. I [Jim Fan] see it differently – with a huge smile on my face. Why are we not happy ... Read more

There are many big questions about the impact of the efficiency gains from the improved DeepSeek methods. Deepseek improved the Reinforcement learning. Deepseek also directly accessed the Nvidia chips. They did NOT use Nvidia CUDA. CUDA was preventing them from doing what they needed to do with the chips. The Deepseek research paper gave a ... Read more

The Opensource DeepSeek R1 model and the distilled local versions are shaking up the AI community. The Deepseek models are the best performing open source models and are highly useful as agents and for other tasks. Here are videos that guide people through using the models. There will soon be many options for Deepseek variants ... Read more

Altimeter Capital analyst and partner puts what Deepseek claims and results into numbers. $6M Training Costs = Plausible IMO Quick math: Training costs ∝ (active params * tokens). DeepSeek v3 (37B params; 14.8T tokens) vs. Llama3.1 (405B params; 15T tokens) = v3 theoretically should be 9% of Llama3.1’s cost. And the disclosed actual figures aligned ... Read more

Langchain used ollama to install Deepseek 14B on a laptop. They used for a local deep researching model. $ ollama pull deepseek-r1:14b $ export TAVILY_API_KEY= $ uvx –refresh –from “langgraph-cli[inmem]” –with-editable . –python 3.11 langgraph dev

Making AI to 10 to 30 times more efficient for AI inference and getting more value from training will increase AI demand. Increased AI efficiency on training and inference will accelerate the improvement and usefulness of AI. Dr Know It All went over the DeepSeek paper and explains how they automated the Reinforcement Learning. AlphaZero ... Read more

DeepSeek just dropped a new open-source multmodal AI model, Janus-Pro-7B. It is MIT opensource license. It’s multimodal (can generate images) and beats OpenAI’s DALL-E 3 and Stable Diffusion across GenEval and DPG-Bench benchmarks. This comes on top of all the R1 hype. Here is the link to the Deepseek Janus 7B Github. NEWS: DeepSeek just ... Read more

Youtuber Ominous Industries provides a training video for setting up your Jetson Nano, installing Pytorch and setting up your own (likely open source LLM). This takes the NVIDIA Jetson Nano to the next level by training a Large Language Model (LLM) completely from scratch. It uses the lightweight and efficient NanoGPT framework, This covers setting ... Read more

Youtuber, Ominous Industries, ran a couple of versions of the DeepSeek R1 1.5B of models running locally on the NVIDIA Jetson Nan. The newly released distilled DeepSeek models were explroed. The DeepSeek R1 1.5B model delivers impressive performance with plenty of room to spare on the Jetson. He shows the installation process, followed by a ... Read more

It is possible to load and run 14 Billion parameter llm AI models on Raspberry Pi5 with 16 GB of memory ($120). However, they can be slow with about 0.6 tokens per second. A 13 billion parameter model can run at 1.36 tokens per second. Improved firmware (better SDRAM timing) improved results.

In March ,2024, Nextbigfuture started covering NANO Nuclear. It was then a startup that had raised over $8 million to develop micro nuclear fission reactors with up to 2 megawatts of power. The reactors will be transported by Semi Trucks. I, Brian Wang of Nextbigfuture, was contacted by the Nano Nuclear team for a correction ... Read more

The Opensource Deepseek R1 AI model is top notch in terms of math, reasoning and coding but it has a lot of bias and restrictions. Here is a video describing some of the examples. It will describe how to use VPNs to get around the China firewall. It provides a Chinese communist part version of ... Read more