SpaceX is reportedly preparing for what could become the largest IPO in history, with a target valuation above $1.75 trillion and plans to raise up to $75 billion. That alone would make it one of the biggest finance stories of the year. But what makes this especially interesting is that the company is no longer just a space business in the traditional sense. In the file, this move is framed as a new kind of public market story where rockets, AI and data infrastructure are combined into one giant company.

That is why this matters far beyond retail investors. For CEOs and finance leaders, SpaceX’s possible listing is a signal of where the market may place the highest value next. It is not only rewarding software anymore. It is rewarding companies that control distribution, infrastructure, compute-heavy ambition and long-term strategic assets. If investors buy into this at scale, it strengthens the idea that the next generation of market leaders will be built around deep integration of physical systems and AI, not just standalone chat products.

A new AI benchmark just dropped – and it made every top model look broken overnight. ARC-AGI-3, released by François Chollet’s ARC Prize Foundation, shows that even the most advanced systems score below 1%, with leaders like Gemini, GPT, and Claude barely registering results. For models that dominate most benchmarks, this kind of reset is rare – and telling.

But this isn’t new for ARC. Every time a new version is released, scores collapse to near zero – and then, surprisingly fast, they rise again. With the previous iteration, labs pushed performance from just a few percent to around 50% in under a year. That pattern raises a deeper question: are models actually learning to reason, or just getting better at optimizing for the test?

To understand why this keeps happening, you need to look at how ARC evolved. The original ARC tests were static puzzles – small grid-based tasks where models had to infer hidden rules. No instructions, no prior examples. ARC-AGI-2 pushed this further, and labs poured resources into optimizing for it, driving scores from near zero to around 50% in less than a year. But much of that progress came from scaling compute and clever search strategies, not necessarily true reasoning.

Most AI conversations today live in extremes – utopia or catastrophe. Anthropic took a different route and asked a simple question: what do people actually want from AI? They gathered answers from 80,000+ users across 159 countries, making this one of the largest qualitative AI studies to date.

The answers are far more grounded than the headlines. The top use case is professional excellence (18.8%) – not replacing work, but removing the mental overload around it. Close behind are personal transformation (13.7%), life management (13.5%), and time freedom (11.1%).

In other words, people are not asking AI to take over their lives – they are asking it to give their lives back.

The newest a16z Top 100 Gen AI Consumer Apps report reveals an important shift: AI is no longer limited to a few standalone chatbots. Instead, it is rapidly becoming a core layer inside the tools people already use every day.

A few years ago, the AI landscape was clearly divided between AI-native tools like ChatGPT or Midjourney and traditional software. That distinction is now disappearing. Popular products such as Canva (Magic Suite), CapCut (AI video editing), and Notion AI have integrated generative AI deeply into their products, and in some cases these AI features are now responsible for a large share of revenue and growth.

One of the most remarkable AI developments this week is not coming from Silicon Valley alone – it is happening in Estonia. OpenAI has partnered with the University of Tartu and Stanford researchers to develop a new framework for measuring how AI actually impacts learning. As part of this collaboration, nearly 20,000 Estonian high school students will participate in a large-scale study exploring how tools like ChatGPT influence learning over time.

For a country of just 1.3 million people, this is a significant moment. Estonia has long been known for its digital society and forward-thinking education system, but this collaboration places the country at the center of global AI education research. Instead of simply adopting AI tools, Estonia is helping shape how the world will measure their real impact on students.

The research goes far beyond traditional exam results. The new framework – called the Learning Outcomes Measurement Suite – studies how AI affects deeper learning behaviors such as motivation, persistence, engagement, metacognition, and knowledge retention. In other words, the focus is not just whether students get better grades, but how AI changes the way people learn.

OpenAI recently published a report about how criminals and state actors try to misuse AI tools.

At first glance, this sounds like another story about AI writing scams or propaganda. But the real insight is something else.

AI is not just being used to write messages. It is being used to run operations more efficiently.

In romance scams, AI helped scammers write emotional messages, translate conversations, create fake documents, and even organize internal reports about victims.