Good morning, it’s Wednesday! Google’s teaming up with Kairos Power to fuel its AI data centers with small nuclear reactors by 2030, aiming for sustainable GPU power—regulatory green lights pending. In other news: OpenAI snags a top Microsoft VP, the U.S. considers tighter chip export rules, and Zyphra and NVIDIA launch a new, high-speed model. Let’s dive in!

Inside Today’s Edition:

1. Top stories 🗞️ 

2. Google Taps Nuclear for AI Power ☢️

3. New “Liquid” Model - Benchmarks Are Useless 📽️ 

4. AI Tools for Writing, Project Management & Digital Avatars 🧰 

5. Help Us Build a Better (Forward) Future 📥️ 
   

🗞️ YOUR DAILY ROLLUP

Top Stories of the Day

• State of AI: This report highlights industry advancements, governance lagging behind development, safety concerns, and Chinese models rise despite sanctions.
  

• Microsoft VP Joins OpenAI: Sebastien Bubeck has joined OpenAI, intensifying the talent race for AGI development. This shift reflects ongoing executive reshuffles.
  

• Zyphra, NVIDIA Launch Zamba2-7B: The hybrid-SSM model outperforms competitors in speed and accuracy, invites collaboration to push AI forward.

• NYT Challenges Perplexity AI Usage: The cease-and-desist sent to Perplexity AI spotlights tensions between the media and AI companies over unlicensed material.
  

• US Weighs AI Chip Export Limits: The US is exploring limits on AI chip exports to certain countries, expanding on existing restrictions targeting China.

• Amazon, Databricks Cut AI Costs: Amazon and Databricks partnered to utilize Amazon’s Trainium chips, offering businesses cheaper AI model solutions.

☝️ POWERED BY LANGTRACE

Monitor, Evaluate & Improve Your LLM Apps

Open source LLM application observability, built on OpenTelemetry standards for seamless integration with tools like Grafana, Datadog, and more. Now featuring Agentic Tracing, DSPy-Specific Tracing, & Prompt Debugging Modes, Langtrace helps you manage the lifecycle of your LLM powered applications. Delivering detailed insights into AI agent workflows, helping you evaluate LLM outputs, while tracing agentic frameworks with precision. Star Langtrace on Github!

☢️ NUCLEAR ENERGY

Google Partners with Kairos Power to Use Nuclear Energy for AI Data Centers by 2030

The Recap: Google has signed an agreement with Kairos Power to deploy small nuclear reactors to meet the growing energy demands of its AI data centers. The first reactor is planned to come online by 2030, with additional reactors to follow by 2035. This partnership highlights the tech industry's increasing shift towards nuclear power as a sustainable energy source for AI infrastructure, though it still requires regulatory approval.

Highlights:

• Google aims to launch the first nuclear reactor with Kairos by 2030, with more reactors planned by 2035.

• No financial details or specific plant locations were disclosed.

• Kairos Power specializes in small reactors cooled by molten fluoride salt, a modern alternative to traditional water-cooling systems.

• Nuclear power offers a low-carbon, reliable energy solution, running 24/7 and supporting decarbonization efforts.

• Global data center energy consumption is expected to more than double by 2030, according to Goldman Sachs.

• US regulators granted Kairos Power the first permit in 50 years for a new type of nuclear reactor, and construction on a demonstration plant has already begun in Tennessee.

• Google’s senior energy director, Michael Terrell, highlighted that the grid needs new electricity sources to support AI's growth, calling this partnership a critical step towards reliable energy solutions.

• The plan awaits approval from the US Nuclear Regulatory Commission and other agencies.

Forward Future Takeaways: This partnership underscores the growing importance of sustainable energy sources like nuclear power to meet the rising energy needs of AI infrastructure. Google joins other tech giants, such as Microsoft and Amazon, who are embracing nuclear energy to power their data centers. While promising, these initiatives face regulatory challenges and public concerns about nuclear waste. If successful, Google’s nuclear strategy could become a benchmark for balancing high energy demand with sustainability. Read the full article here.

🛰️ NEWS

Looking Forward: More Headlines

• Anthropic CEO's AI Vision: CEO envisions AI transforming society positively soon, emphasizing risk management to shape a beneficial future.

• Evaluating ML Agents: MLE-bench assesses AI agents on 75 ML engineering challenges, comparing them against human baselines.

• OpenAI’s Meta Prompt: OpenAI's meta-prompt focuses on task-oriented performance, contrasting Anthropic's conversational, collaborative approach.

• First GPU’s Impact: NVIDIA's GeForce 256 revolutionized gaming and became a catalyst for AI advancements through parallel processing.

• Luxurious Space Station Unveiled: Vast reveals Haven-1, a modern commercial space station with luxury amenities, aiming for a 2025 launch.

• AI Finds 70,000 Viruses: AI-driven analysis uncovered 70,500 new RNA viruses, advancing understanding of viral diversity and evolution.

✌️ POWERED BY MAMMOUTH AI

Access the Best of GenAI for $10/Month

Get access to the best LLMs (GPT-o1, Claude 3.5, Llama 3.1, chatGPT-4o, Gemini Pro, Mistral) and the best AI generated images (Flux.1 Pro, Midjourney, SD3, Dall-E) in one place for just $10 per month. Enjoy on mammouth.ai.

🧰 TOOLBOX

Top AI Tools for Writing, Project Management & Digital Avatars

• AI Tool for Automating Writing Task: Sloap automates title, description, and tag generation to streamline writing tasks. It enhances productivity by handling repetitive tasks, freeing users to focus on creativity.

• Autonomous Project Management Tool: Height leverages an advanced reasoning engine to automate routine tasks and provide contextual insights, enabling teams to focus on key work.

• AI Studio for Lifelike Digital Avatars: D-ID’s Creative Reality™ Studio uses AI to produce videos with realistic digital avatars. It empowers creators to generate personalized content for marketing and more.

• Seamless Development Query Solution: Unblocked delivers quick, accurate answers by integrating with existing tools and understanding codebases. It aids developers with real-time documentation.

📽️ VIDEO

New “Liquid” Model - Benchmarks Are Useless

Liquid AI's latest generative AI model, the Liquid Foundation Model (LFM), breaks away from traditional Transformer architecture and comes in three sizes—1B, 3B, and a 40B parameter mixture of experts model. While LFM performs well in benchmarking, especially in memory efficiency and handling long context inputs, the model struggled with practical reasoning and logic tests, failing in areas like text counting, basic logic puzzles, and moral reasoning. Watch the full video below 👇️ 

🗒️ FEEDBACK

Help Us Get Better

🤠 THE DAILY BYTE

AI in the EU: More Red Tape Than Code

🌐 CONNECT

Stay in the Know

Follow us on X for quick daily updates and bite-sized content.
Subscribe to our YouTube channel for in-depth technical analysis.

Prefer using an RSS feed? Add Forward Future to your feed here.

Thanks for reading today’s newsletter. See you next time!

Forward Future Team
🧑‍🚀 👨‍🚀 👩‍🚀 👩‍🚀