NVIDIA unveils Earth-2 models while Jensen Huang projects grid strain

Key Takeaways

• NVIDIA launched its Earth-2 family of open AI models and tools for weather and climate prediction on January 26, 2026.
• The Earth-2 platform includes Atlas for 15-day forecasts, StormScope for short-term local storm predictions, and HealDA for fast data assimilation.
• NVIDIA CEO Jensen Huang warned that AI requires a massive infrastructure buildout, projecting US data center power demand to exceed 150 gigawatts by 2028.
• Georgia is considering a statewide moratorium on new AI data centers due to concerns over energy and water consumption.
• Utilities face significant challenges, including staffing shortages and grid limitations, to meet the rising electricity demands from AI data centers.
• Bristol Myers Squibb uses AI, supported by NVIDIA DGX SuperPODs, to accelerate clinical trials and optimize workforce talent.
• AI is improving ALS patient care by enhancing disease classification, diagnosis, treatment, and outcome prediction, as highlighted by Dr. Crystal Yeo.
• AI models can distinguish between healthy individuals and those with ALS or myopathy using electromyography signals.
• Power availability, not just chips, is now a critical factor determining the success of AI markets, according to Jensen Huang.
• NoteGPT is an AI-powered tool that summarizes web articles, YouTube videos, and documents, offering timestamped summaries and a chat assistant.

Nvidia launches AI for faster weather forecasts

Nvidia introduced new AI models for weather forecasting at the American Meteorological Society's annual meeting in Houston. These "Earth-2" models aim to replace expensive, slow traditional simulations with faster, cheaper AI versions that match or beat their accuracy. Mike Pritchard from Nvidia noted the insurance industry can use these models to predict extreme events like floods much faster, running 10,000-member forecasts 1,000 times quicker. The models include tools for 15-day forecasts, six-hour severe storm predictions for the US, and integrating various weather sensor data.

NVIDIA unveils Earth-2 open AI weather models

NVIDIA launched its Earth-2 family of open AI models and tools for weather and climate prediction on January 26, 2026, at the American Meteorological Society's Annual Meeting. These models make AI weather forecasting accessible globally, from processing data to generating 15-day global forecasts or local storm predictions. The Earth-2 Nowcasting model uses generative AI and satellite data to forecast storms, while Earth-2 Medium Range, powered by Atlas, offers high-accuracy 15-day forecasts for over 70 variables. Earth-2 Global Data Assimilation, using HealDA, creates initial atmospheric conditions in seconds on GPUs. This open platform helps various organizations, including weather services and energy companies, save time and costs by using AI instead of traditional supercomputer models.

NVIDIA Earth-2 makes AI weather forecasting open

NVIDIA introduced Earth-2, the world's first fully open and accelerated AI weather stack, making climate science more accessible. This platform includes three new models: Atlas for 15-day forecasts, StormScope for short-term local storm predictions, and HealDA for fast data assimilation. Mike Pritchard from NVIDIA stated they provide foundational tools for nations and companies to build their own weather systems, emphasizing national security. StormScope is notable for being the first AI model to outperform traditional methods for short-term precipitation forecasts, generating predictions in minutes. Earth-2 is already helping companies like TotalEnergies and GCL predict solar and wind energy.

Error loading article on AI and universities

This article could not be displayed due to a loading error. The page reported a failure to load a specific chunk of JavaScript. Therefore, no content was available to summarize regarding universities and artificial intelligence.

Bristol Myers Squibb uses AI for trials and talent

Bristol Myers Squibb (BMS) uses artificial intelligence in two main areas to improve its operations. First, AI helps speed up clinical trial operations and patient enrollment by matching patients to trials in real time. This addresses past issues with slow, manual processes and missed patient opportunities. Second, BMS uses large language models to optimize workforce talent, matching employee skills to career development opportunities. The company partnered with Accenture to create Workbench, an AI-driven clinical trial accelerator, and with NVIDIA for powerful computing infrastructure like the DGX SuperPOD. This technology helps train AI models on clinical images, enhancing research speed and accuracy.

Georgia considers banning new AI data centers

Georgia is leading efforts to ban new data centers, which power America's AI growth, due to concerns about energy and water use. State Democratic legislator Ruwa Romman introduced a bill for a statewide moratorium until March next year, allowing time to create regulations. The state's public service commission plans a 10-gigawatt power increase, but the Atlanta area alone needs 17 times more power by 2030. Critics like Charles Hua point out that Georgia Power profits from new investments, leading to higher electricity rates for residents. Romman's bill also aims to give Georgians time to vote on Public Service Commission seats, potentially shifting the commission's stance on approving power demands for data centers.

AI helps improve ALS patient care

Dr. Crystal Yeo, a neurologist from Singapore's National Neuroscience Institute, discussed how artificial intelligence can advance ALS patient care. She highlighted AI's role in improving disease classification, diagnosis, treatment, monitoring, and predicting outcomes for neuromuscular disorders. AI models can already distinguish between healthy individuals and those with ALS or myopathy using electromyography signals. Dr. Yeo emphasized the need for better patient grouping and AI models that explain their decisions to overcome challenges like disease variation. She also stressed the importance of teamwork across data science, research, and clinical care to make these advancements useful for patients.

Utilities must prepare for AI energy demands

The demand for electricity in the US will rise significantly by 2030 and 2050, mainly due to new AI data centers. This creates big challenges for utility companies, including staffing shortages and grid limitations. A single large data center can use as much power as 400,000 electric cars, requiring major overhauls to existing power systems. Building new power plants takes years, so utilities might need to restart old ones or find ways to be more efficient. Companies must assess if they have enough skilled engineers and consider hiring outside experts to manage these complex projects.

Nvidia CEO warns AI needs massive power

Nvidia CEO Jensen Huang warned Wall Street that artificial intelligence requires a massive infrastructure buildout, calling it the largest in human history. Speaking at the World Economic Forum, Huang outlined a "five-layer cake" of AI, emphasizing that energy and power form the crucial base layer. He noted that US data center power demand could jump from 80 gigawatts in 2025 to over 150 gigawatts by 2028. Huang stressed that power availability, not just chips, now determines the success of AI markets. Utilities are already seeing significantly increased load projections, and even Big Tech is using the bond market to fund this energy-intensive expansion.

NoteGPT AI summarizes YouTube videos and articles

NoteGPT is an AI-powered productivity tool designed to create concise summaries and notes from various content types. It can summarize web articles, YouTube videos, and uploaded documents like PDFs. A key feature is its ability to transcribe YouTube audio and provide timestamped summaries, helping users quickly grasp main points. NoteGPT also includes a context-aware AI chat assistant that allows users to ask specific questions about the summarized content. While it excels at information digestion and has an easy-to-use interface, its summary quality can miss nuances in complex topics, and its free plan is limited.

Sources

• Nvidia unveils AI models for faster, cheaper weather forecasts
• NVIDIA Launches Earth-2 Family of Open Models — the World’s First Fully Open, Accelerated Set of Models and Tools for AI Weather
• NVIDIA Revolutionizes Climate Tech with ‘Earth-2’: The World’s First Fully Open Accelerated AI Weather Stack
• Universities and the Challenge of Artificial Intelligence
• Artificial Intelligence at Bristol Myers Squibb - Two Use Cases - Emerj Artificial Intelligence Research
• Georgia leads push to ban datatcenters used to power America’s AI boom
• Integrating AI and Translational Science to Advance ALS Care: Crystal Yeo, MD, PhD | NeurologyLive - Clinical Neurology News and Neurology Expert Insights
• How Utilities Can Prepare for the AI-Driven Energy Surge
• Nvidia CEO Jensen Huang gives AI warning Wall Street can’t ignore
• Recensione di NoteGPT (ottobre 2026): Riassunti AI per YouTube e articoli