In Today’s Issue:

🛑 Progressive lawmakers are pushing for a temporary moratorium on new AI data centers

🌍 Google Earth is officially transforming from a visualization tool into AI-powered geospatial intelligence platform.

🤖 NVIDIA is betting $26 billion on open AI models

🧠 Lithium could be the brain's hidden shield against Alzheimer's!

✨ And more AI goodness…

Dear Readers,

NVIDIA just dropped a 120-billion-parameter model that only fires 12 billion neurons per call - and it might be the most consequential open-source release of 2026 so far. Nemotron 3 Super isn't just another benchmark-topper; it's a blueprint for building sophisticated AI agents without handing the keys to a proprietary provider, and today we break down exactly why that matters for developers, startups, and anyone betting on the agent economy.

But that's not all on your plate: progressive lawmakers in Washington are pushing a moratorium on new data centers, Google is turning Earth into an AI-powered geospatial command center, and NVIDIA is backing its model ambitions with a staggering $26 billion war chest - signaling it wants to be the lab, not just the chip supplier.

We also dig into a fascinating piece of science: Harvard researchers found that lithium — yes, the element — may be the brain's hidden shield against Alzheimer's, opening a prevention pathway that could cost pennies a day.

And speaking of NVIDIA: next week, GTC 2026 kicks off in San Jose (March 16–19), with Jensen Huang taking the stage at SAP Center in front of 30,000 attendees from 190 countries — and Superintelligence will be there live, bringing you on-the-ground coverage of the announcements, demos, and hallway conversations that don't make it into press releases.

Grab your coffee and dig in.

You can also watch the livestream from home. You can find out everything else about GTC 2026 at the following link:

All the best,

Kim Isenberg

Some progressive U.S. lawmakers, led by Bernie Sanders, are gaining support for a temporary moratorium on new AI data centers due to concerns about energy use, water consumption, and environmental impact. Supporters argue the pause would give regulators time to catch up with the rapid expansion - potentially thousands of new facilities nationwide - while critics warn it could weaken the U.S. in the global AI race with China. Polling shows 50% of voters support more data centers, but support drops to 36% if built within three miles of homes, highlighting growing local tensions.

Elon Musk: “I believe the economy will 10x in the next 10 years unless we get WW3.”

Editor-in-Chief Kim Isenberg took the opportunity to attend the opening of the GoogleAI research center in Berlin and to speak with exciting people on site!

NVIDIA just dropped a model that could reshape how we build AI agents. Nemotron 3 Super, released on March 11, packs 120 billion parameters into a hybrid architecture that only activates 12 billion during inference - making it radically efficient for its size. The secret sauce: a Latent Mixture-of-Experts design combined with Mamba-2 layers and Transformer attention, enabling the model to consult four times as many expert modules at the same computational cost as standard approaches. Add a native one-million-token context window, and you get an agent brain that can hold entire codebases or thousands of report pages in memory without losing track of its goals.

Multi-agent systems typically generate up to 15x more tokens than regular chatbots, creating a “context explosion” that causes goal drift over long tasks. Super tackles this head-on with 5x higher throughput than its predecessor and built-in Multi-Token Prediction for speculative decoding, delivering up to 3x wall-clock speedups on structured tasks like code generation. Trained on 25 trillion tokens using NVIDIA’s 4-bit NVFP4 format from day one - not quantized after the fact - the model ships in three precision levels: BF16 for maximum accuracy, FP8 as a balanced middle ground, and NVFP4 for peak throughput on Blackwell GPUs. Already available through Hugging Face, Perplexity, and a dozen cloud partners, NVIDIA is open-sourcing the complete stack: weights, 10 trillion curated pretraining tokens, training recipes, and 15 reinforcement learning environments. Could this be the efficiency breakthrough that finally makes sophisticated multi-agent systems practical at scale?

Why it matters: Nemotron 3 Super proves that open-source models can rival proprietary systems in agentic reasoning while being radically more efficient. By releasing the full training stack alongside the weights, NVIDIA is handing developers the blueprint to build, customize, and deploy sophisticated AI agents on their own infrastructure - a significant shift in who gets to participate in the agent economy.

Sources:

🔗 https://developer.nvidia.com/blog/introducing-nemotron-3-super-an-open-hybrid-mamba-transformer-moe-for-agentic-reasoning/

🔗 https://huggingface.co/nvidia/NVIDIA-Nemotron-3-Super-120B-A12B-FP8?ncid=ref-inor-231713

Markets don’t wait — and year-end waits even less.

In the final stretch, money rotates, funds window-dress, tax-loss selling meets bottom-fishing, and “Santa Rally” chatter turns into real tape. Most people notice after the move.

Elite Trade Club is your morning shortcut: a curated selection of the setups that still matter this year — the headlines that move stocks, catalysts on deck, and where smart money is positioning before New Year’s. One read. Five minutes. Actionable clarity.

If you want to start 2026 from a stronger spot, finish 2025 prepared. Join 200K+ traders who open our premarket briefing, place their plan, and let the open come to them.

Get Year-End Briefing

_{By joining, you’ll receive Elite Trade Club emails and select partner insights. See Privacy Policy.}

Despite the tech-bubble narrative that “ChatGPT is losing,” the data shows the opposite: ChatGPT still dominates the AI market with ~900M weekly users, far higher engagement, and significantly more traffic than Gemini or Claude. However, it’s losing market share to Gemini and Claude.

Harvard scientists may have found Alzheimer’s missing puzzle piece - and it was hiding on the periodic table all along. A landmark study published in Nature by Dr. Bruce Yankner’s team revealed that lithium occurs naturally in the brain and that its depletion is among the earliest molecular changes preceding Alzheimer’s. The researchers analyzed 27 metals in brain tissue from nearly 400 individuals and found lithium was the only element significantly reduced in people with mild cognitive impairment, the stage before full-blown Alzheimer’s. The culprit: amyloid plaques act as molecular sponges, stealing lithium ions and starving neurons exactly when they need protection most.

In mice, a lithium-deficient diet dramatically accelerated plaque formation, tau tangles, and cognitive decline. The hopeful twist: a compound called lithium orotate reversed memory loss at roughly one-thousandth of the psychiatric dose - with zero toxicity across nearly their entire lifespan. Unlike traditional lithium carbonate, lithium orotate evades amyloid capture and reaches vulnerable neurons. Human trials are now being prepared. Epidemiological data from Japan and Texas already link naturally higher lithium in drinking water to lower dementia rates. If confirmed in humans, Alzheimer’s prevention could shift from chasing toxic proteins to restoring what the brain quietly relied on all along.

This research reframes Alzheimer’s not just as a disease of toxic protein accumulation, but as a state of critical micronutrient deficiency - opening a fundamentally new avenue for prevention. If lithium orotate proves safe and effective in human trials, it could become one of the most accessible and affordable neuroprotective interventions in history, available over the counter for pennies a day.

Be the smartest person in the room. 1440 navigates 100+ sources to deliver a comprehensive, unbiased news roundup — politics, business, culture, and more — in a quick, 5-minute read. Completely free, completely factual.

Sign up now!