Become the go-to AI expert in 30 days

AI keeps coming up at work, but you still don't get it?

That's exactly why 1M+ professionals working at Google, Meta, and OpenAI read Superhuman AI daily.

Here's what you get:

• Daily AI news that matters for your career - Filtered from 1000s of sources so you know what affects your industry.

• Step-by-step tutorials you can use immediately - Real prompts and workflows that solve actual business problems.

• New AI tools tested and reviewed - We try everything to deliver tools that drive real results.

• All in just 3 minutes a day

Join 1M+ pros

Today:

• Why Google Might Quietly Win the AI Race by 2035

• SoftBank Exits Nvidia, Pours $5.8B Into AI Projects

• Yann LeCun to Exit Meta, Launch AI Startup Focused on World Models

• Anthropic Aims to Beat OpenAI with Cheaper, Smarter AI Infrastructure

• Microsoft and Google Pledge $16B+ for AI Infrastructure in Europe

no one sees it coming (except Google)

Google’s research hints it could solve AI’s hurdles. Papers show “nested learning” for continuous learning, evidence that transformers build “maps” for reasoning, and a Gemma biology model that found a new cancer pathway, backing scaling laws. 

Google also pursues energy with Project Suncatcher with a prototype slated for 2027 and possible cost parity around 2035. On chips, seventh-gen TPUs offer efficient training and inference and are rented via cloud. Together, these moves could unlock profit from drug discovery and make Google the long-run winner.

WATCH THE VIDEO ON YOUTUBE

SoftBank Exits Nvidia, Pours $5.8B Into AI Projects

SoftBank has sold all its shares in Nvidia for $5.83 billion to fund its growing number of AI projects, including partnerships with OpenAI and Oracle. The move highlights concerns over whether massive tech investments in artificial intelligence will truly pay off.

KEY POINTS

• SoftBank sold its full Nvidia stake to raise $5.83 billion for AI investments.

• Masayoshi Son is reallocating funds into major AI projects like data centers and robotics.

• Investors are worried about whether trillion-dollar AI spending by tech giants will bring real returns.

Why it matters

This shows how serious big players are about artificial intelligence—so much that they’re selling valuable stocks to bet on it. But it also raises a red flag: Will all this money spent on AI actually lead to useful results, or is it a risky gamble?

Yann LeCun to Exit Meta, Launch AI Startup Focused on World Models

Yann LeCun, Meta’s chief AI scientist and Turing Award winner, plans to leave the company to launch his own startup focused on “world models” — AI systems that understand and predict the world. His exit highlights growing tensions inside Meta’s shifting AI strategy.

KEY POINTS

• Yann LeCun is leaving Meta to start a new AI company centered on world models.

• Meta’s AI division is in flux, with internal chaos and a shift toward short-term results under the Meta Superintelligence Labs (MSL).

• LeCun is skeptical of AI hype, especially large language models, and has publicly questioned their real-world intelligence.

Why it matters

LeCun helped shape today’s AI but is now walking away from a tech giant to pursue his vision elsewhere. That says a lot. It shows cracks inside Meta and raises questions about whether today’s AI race is heading in the right direction—or just chasing hype.

Anthropic Aims to Beat OpenAI with Cheaper, Smarter AI Infrastructure

Anthropic believes it can run AI far more efficiently than OpenAI, projecting much lower computing costs and earlier profitability. By leaning on cheaper chips and high enterprise demand, it expects to turn cash-flow positive by 2027—two years ahead of OpenAI.

KEY POINTS

• Cost Edge: Anthropic projects $6B compute spend in 2025 vs. OpenAI’s $15B, widening to $27B vs. $111B by 2028.

• Profit Timeline: Anthropic expects to be cash-flow positive in 2027; OpenAI not until at least 2029.

• Revenue Model: 80% of Anthropic’s income comes from paid API usage, unlike OpenAI’s free-heavy ChatGPT user base.

Why it matters

This shows Anthropic’s strategy is all about running lean—using cheaper chips, spending less, and focusing on paying users. It could make Anthropic more stable and profitable sooner, while OpenAI takes a bigger gamble on huge spending and future monetization.

🧠RESEARCH

Thinking with Video: Video Generation as a Promising Multimodal Reasoning Paradigm

The paper introduces “Thinking with Video,” a new way for AI to reason by using videos instead of just images or text. This method helps AI better understand changes over time. Their model, Sora-2, performs well on both visual and math tasks, showing video-based reasoning could unify how AI thinks.

V-Thinker: Interactive Thinking with Images

V-Thinker is a new AI assistant that helps models think more deeply using images. It learns through trial and error, improves over time, and focuses on understanding complex visual tasks. Using custom training and a new benchmark, it beats other models in image-based reasoning, pushing interactive visual thinking forward.

HaluMem: Evaluating Hallucinations in Memory Systems of Agents

HaluMem is a new benchmark that tests how and when AI memory systems make things up—like wrong facts or missing details. By breaking memory use into stages (saving, updating, answering), the study shows most errors start early and spread. It calls for better tools to make AI memory more trustworthy.

🛠️TOP TOOLS

Each listing includes a hands-on tutorial so you can get started right away, whether you’re a beginner or a pro.

AI Code Mentor – Code Optimization, Refactoring and Review - browser-based assistant that explains snippets of code in plain language

AI Code Converter – Simple, Free Code Converter & Generator - web-based tool that translates code between programming languages

Afforai – Research, Cite & Write with AI - AI‑powered workspace that brings research, reference management, PDF annotation, and academic writing

📲SOCIAL MEDIA

🗞️MORE NEWS

• Microsoft and Google are investing over $16 billion to build AI infrastructure in Europe. Microsoft will build a major data hub in Portugal, while Google expands offices and data centers across Germany through 2029.

• Baidu launched an open-source AI model that it claims beats GPT-5 and Gemini in visual reasoning while using fewer resources. The model features dynamic image analysis, runs on a single GPU, and allows commercial use.

• A German court ruled that ChatGPT violated copyright by training on protected song lyrics. OpenAI must pay damages, setting a major precedent in Europe for AI accountability and the protection of creative content.

• Google launched Private AI Compute, a cloud system that powers advanced Gemini AI features while keeping personal data secure and private. It combines cloud performance with strict privacy controls, mimicking on-device protection standards.

• England will assign AI-generated attendance targets to every school to boost post-pandemic student attendance. Unions criticize the move, saying it adds pressure without addressing deeper causes behind absences, like poverty and health issues.

• Israeli AI startup Wonderful raised $100 million, boosting its valuation to $700 million. Its multilingual AI agents serve businesses across chat, email, and voice. The company plans Asia-Pacific expansion and targets $10 million in revenue.

• A Dartmouth study shows AI teaching tools like NeuroBot TA can personalize learning and earn student trust by using only vetted course materials. The system helps medical students study efficiently and may aid global education access.