Category: Uncategorized

A Foundation Model is a massive AI model trained on a vast amount of data — text, images, code — that can be adapted to perform a wide range of tasks, from writing emails to diagnosing diseases.

Hey Common Folks!

In our last article, we explored Generative AI — AI that can create brand new content like essays, images, music, and code. We saw it transforming customer support, education, content creation, and software development.

But here’s a question that article left open: what’s actually powering all of that?

When ChatGPT writes your email, when Claude explains your kid’s homework, when Gemini summarizes a research paper — they’re all running on the same type of technology underneath. That technology is called a Foundation Model.

The Old Way vs. The New Way

In traditional AI (the old way), if you wanted to translate English to French, you built a “Translation Model.” If you wanted to summarize text, you built a “Summarization Model.” If you wanted to detect spam, you built a “Spam Model.”

One model, one job. Every new task meant starting from scratch.

Foundation Models changed the game completely. They are like a Swiss Army Knife — one single tool that can perform hundreds of different tasks, from writing code to composing poetry to analyzing legal contracts.

The Analogy: The “Super-Student” in the Library

Remember from our earlier article on What is a Model — we compared an AI model to a student who finished studying and walks into the exam with knowledge in their head?

Now imagine two types of students:

• Traditional AI (The Specialist): This student only studied one book: “How to Repair a Bicycle.” Ask them to fix a bike? Perfect. Ask them to write an essay on history? They have no idea. They are specialized.

• Foundation Model (The Generalist): This student has read almost every book in the library. History, math, coding, poetry, mechanics, medicine. Because they’ve seen so much, they’ve learned general patterns about how the world works.

Now, you can ask this “Super-Student” to fix a bike, or write a poem, or solve a math problem, or draft a legal brief. They have a broad “foundation” of knowledge that allows them to adapt to almost any request.

That’s why they’re called Foundation Models — they serve as the base upon which everything else is built. Just like you lay a concrete foundation before building a house, a hospital, or a skyscraper.

The Framework: Builders vs. Users

Understanding Foundation Models becomes much easier when you split the world into two groups: the people who make the engine, and the people who drive the car.

1. The Builder Perspective (Making the Brain)

These are the engineers at companies like OpenAI, Anthropic, Google, or Meta. They take massive amounts of raw materials — text from the internet, books, code repositories, images — and use sophisticated processes to train these models.

• The Goal: To create a model that learns general patterns about language, logic, and the world. It’s like feeding the “student” all those books.

• The Result: A Foundation Model (like GPT-4o, Claude, Gemini, or Llama).

2. The User Perspective (Putting the Brain to Work)

This is where most businesses and developers sit. They don’t need to know how to build the brain from scratch. They just need to know how to use it.

Think of the Foundation Model as a powerful Engine:

• One user might take that engine and put it into a Race Car (a chatbot for customer service).

• Another user might put it into a Truck (a tool to summarize legal documents).

• Another might put it into a Boat (an app that generates marketing copy).

• Another might put it into an Ambulance (an AI assistant helping doctors with diagnoses).

You don’t need to know how to build the engine. You just use its capabilities to solve your specific problem.

This is exactly what’s happening in the real world right now. Companies like Canva, Notion, Duolingo, and Khan Academy aren’t building their own Foundation Models — they’re plugging into existing ones (GPT, Claude, Gemini) and building products on top.

Types of Foundation Models

While Large Language Models (LLMs) are the most famous type, Foundation Models are expanding into new territories:

1. Large Language Models (LLMs): Trained on text. Good for writing, summarizing, reasoning, and coding. Examples: GPT-4o, Claude, Llama, Gemini.

2. Large Multimodal Models (LMMs): These can understand not just text, but also images, audio, and video. When you upload a photo to ChatGPT and ask “what’s in this image?” — that’s a multimodal model at work. Examples: GPT-4o (handles text + images + audio), Gemini (text + images + video).

3. Image Generation Models: Trained on images paired with descriptions. They create visuals from text prompts. Examples: DALL-E 3, Midjourney, Stable Diffusion.

4. Code Models: Specialized for understanding and generating software code. Examples: Claude (Anthropic’s model powers Claude Code), GitHub Copilot (powered by OpenAI models).

All of these are Foundation Models — massive, general-purpose brains that can be adapted to specific jobs.

Why This Matters for You

You might be thinking: “Okay, but I’m not building AI. Why should I care about Foundation Models?”

Three reasons:

1. It explains why AI suddenly got good at everything. Before Foundation Models, AI was narrow — good at one thing, useless at the rest. Foundation Models are why your AI assistant can write an email and explain physics and debug code and plan your vacation. One brain, many skills.

2. It’s why the AI race is so expensive. Training a Foundation Model costs tens of millions to hundreds of millions of dollars. That’s why only a handful of companies (OpenAI, Anthropic, Google, Meta) can afford to build them. Everyone else builds on top of them.

3. It’s the reason AI will keep getting more useful. As Foundation Models get better, every product built on top of them gets better automatically. When GPT improves, every app using GPT improves. That’s the power of a shared foundation.

The Takeaway

A Foundation Model is a general-purpose AI brain:

• It is the engine inside the car.

• It is the student who read every book in the library.

• It is the Swiss Army Knife of the digital world.

It shifted AI from “specialized tools that do one thing” to “general intelligence that can be adapted to almost anything.”

And here’s the exciting part: we’re still early. The Foundation Models of 2026 are dramatically more capable than those of 2023. The ones coming in 2027 and beyond will make today’s look primitive. Understanding this technology now means you won’t be caught off guard as it transforms more of the world around you.

Coming Up

Now that you understand what a Foundation Model is — this massive, general-purpose brain — you’ve probably noticed we keep mentioning one name more than any other: GPT. GPT-3, GPT-4, ChatGPT. But what do those three letters actually stand for? And why did this particular approach become the dominant one? In our next article, we’ll decode the most famous acronym in AI and break down exactly how GPT works, letter by letter.

AI for Common Folks — Making AI understandable, one concept at a time.

Subscribe now

Leave a comment

Good morning, investors are lining up to value Anthropic at $800 billion while OpenAI’s own backers question its $852 billion price tag, OpenAI just launched a cybersecurity-focused model to answer Anthropic’s Mythos, and Uber committed $10 billion to robotaxis in its biggest strategic pivot ever. Here’s what happened 👇

1. Anthropic’s Valuation Could Double to $800 Billion as OpenAI Faces Investor Doubt

Venture capital firms have approached Anthropic with offers to invest at valuations as high as $800 billion, more than double the $380 billion it raised at in February. Anthropic has so far resisted these overtures, according to Bloomberg. The company’s run-rate revenue now surpasses $30 billion, up from roughly $9 billion at the end of 2025, driven by surging demand for Claude and the buzz around its frontier Mythos model.

Meanwhile, the Financial Times reported that some of OpenAI’s own backers are questioning its $852 billion valuation as the company shifts its strategy toward the enterprise market to compete with Anthropic. The contrast is striking: one company’s investors are racing to get in. The other company’s investors are asking if they overpaid.

Why it matters: Six months ago, OpenAI was the undisputed leader in AI. Now Anthropic is growing revenue at a pace that could close the gap faster than anyone expected. If you use ChatGPT or Claude at work, you are watching a competitive shift that will directly affect the tools, pricing, and features available to you.

Source: Reuters | Source: Reuters

2. OpenAI Launches GPT-5.4-Cyber to Counter Anthropic’s Mythos

OpenAI unveiled GPT-5.4-Cyber, a variant of its latest flagship model fine-tuned specifically for defensive cybersecurity work. The release comes exactly one week after Anthropic announced Mythos, which has already found “thousands” of major vulnerabilities in operating systems, browsers, and other software. GPT-5.4-Cyber will initially be available only to vetted security vendors, organizations, and researchers through OpenAI’s expanded Trusted Access for Cyber (TAC) program.

The highest-tier TAC users will get access to the model with fewer restrictions on sensitive cybersecurity tasks like vulnerability research and analysis. OpenAI is also opening the program to thousands of individual defenders and hundreds of security teams.

Why it matters: The cybersecurity AI race is now a two-horse competition. Both OpenAI and Anthropic are building models specifically designed to find software vulnerabilities before attackers do. For companies, this means AI-powered security tools are about to get significantly more capable. For everyone else, it means the software you use every day is about to get tested by AI systems that can find flaws humans have missed for years.

Source: Reuters

3. Uber Commits $10 Billion to Robotaxis, Breaks Its Own Business Model

Uber has committed more than $10 billion to buying thousands of autonomous vehicles and taking equity stakes in their developers, according to the Financial Times. This is a fundamental break from the “gig economy” model that built the company. Uber is positioning itself as a marketplace for multiple robotaxi operators, partnering with Baidu, Rivian, and Lucid, and plans to launch robotaxi services in at least 28 cities by 2028.

The deals include roughly $2.5 billion in equity stakes and over $7.5 billion in fleet purchases over the next few years, contingent on partners hitting deployment milestones.

Why it matters: The company that defined ride-sharing is betting its future on removing drivers entirely. If you use Uber, you could be hailing a driverless car within two years depending on your city. This is also a signal that the robotaxi market has crossed from “maybe someday” to “we need to own this now” for the biggest players in transportation.

Source: Reuters

4. Snap Cuts 1,000 Jobs, Says AI Now Writes 65% of Its Code

Snapchat’s parent company is laying off about 1,000 employees, 16% of its full-time staff, and closing over 300 open positions. The company said AI is now generating more than 65% of new code at Snap, enabling it to operate with smaller teams. CEO Evan Spiegel expects the cuts to save more than $500 million in annualized expenses by the second half of 2026.

Snap is not alone. More than 80 tech companies have cut roughly 71,440 jobs so far this year, according to Layoffs.fyi, as AI adoption accelerates across the industry.

Why it matters: The stat to pay attention to is not the layoff count. It is the 65% number. When a major tech company publicly says AI is writing two-thirds of its new code, that is a signal about where software development is heading across every industry. The question is no longer whether AI will change the job market. It is how fast.

Source: Reuters

Quick Hits

• Maine became the first US state to pass a moratorium on large data centers, freezing approvals for facilities requiring more than 20 megawatts of power until October 2027. Eleven other states are weighing similar legislation. Source: Reuters

• Federal agencies are quietly sidestepping Trump’s ban on Anthropic to test its Mythos model. The Commerce Department’s Center for AI Standards is actively testing Mythos’ capabilities, and staff on at least three congressional committees have held or requested briefings. Source: Reuters

• Jane Street signed a $6 billion AI cloud computing deal with CoreWeave and boosted its equity stake in the company, one of the largest single cloud contracts announced this year. Source: Reuters

That’s it for today. The AI industry is splitting into two clear lanes: one where the biggest companies race to build the most powerful models, and another where everyone else figures out what those models mean for their workers, their cities, and their energy bills.

Forward this to someone who needs to stay in the loop.

Subscribe now

Leave a comment

Generative AI is artificial intelligence that creates brand new content that didn’t exist before, like writing an essay, drawing a picture, composing music, or writing computer code, by learning patterns from mountains of existing data and producing something new that feels like a human made it.

Hey Common Folks!

We break down AI so it makes sense to real people. If you’ve been following along, our last article explored Predictive Modeling, how AI uses historical data to make educated guesses about the future. That’s AI looking backward to predict forward.

Today we’re flipping the script. What if AI could look at everything it has learned and create something entirely new?

That’s Generative AI. And it’s the reason AI went from a behind-the-scenes tool to something your parents, your boss, and your neighbor are all talking about.

Why This Feels Different from Everything Before

Think about what makes humans special. Our ability to create new things, right? For decades, people said this was the one thing AI would never be able to do.

Before Generative AI, artificial intelligence was mainly used for prediction and classification:

• Will this customer buy our product? (prediction)

• Is this email spam or not? (classification)

• Which movies should we recommend? (recommendation)

Useful, but limited. Now, Generative AI can write a poem, design a logo, generate a 3D model, compose music, or write working software. The creative barrier has been broken.

In fact, the images you see in this article and even the majority of this writing style was created with the help of AI tools. We’re using the very technology we’re explaining.

Four Ways Generative AI Is Already Changing the World

1. Customer Support That Doesn’t Make You Pull Your Hair Out

Remember the last time you needed help and had to wade through an endless phone tree or wait hours for a response?

Companies used to need huge call centers with dozens of employees handling problems. Expensive and frustrating for everyone.

Now, AI-powered chatbots handle the first level of support. When you contact customer service for a food delivery app or your internet provider, your initial questions are likely answered by an AI that understands what you’re asking and gives helpful responses.

The result? Companies reduce their support staff from 10 people to 2-3, while actually improving response times. The AI handles common questions. Human agents focus on the complex issues that really need their attention.

2. Content Creation That’s Indistinguishable from Human Work

“AI can’t be creative” was the mantra for years. That ship has sailed.

Today, if you read an article online, you often cannot tell if a human or an AI wrote it. The quality has improved that dramatically.

This extends beyond writing to image creation, video editing, music composition, and more. Artists and creators aren’t being replaced, but they’re increasingly working alongside AI tools that speed up their workflow.

For everyday people, this means you can generate professional-quality content without years of specialized training. Need a business proposal? A birthday card design? A custom bedtime story for your kids? Generative AI can help create it in minutes.

3. Education That Adapts to How You Learn

Generative AI is transforming education by providing 24/7 personalized learning support. Stuck on a complex topic? AI can explain concepts in multiple ways until you understand.

Tools like ChatGPT, Claude, and Gemini are already being used by students worldwide as study partners. Universities like Northeastern and the London School of Economics have integrated AI tutoring into their programs. Some of these tools use Socratic questioning, asking you guiding questions instead of handing you the answer, helping you actually think instead of just copy.

This newsletter itself is an example. We use AI tools every day to research, draft, and refine explanations so that complex topics reach you in plain language.

4. Software Development That’s Accessible to Almost Everyone

Coding used to be a highly specialized skill requiring years of training. Generative AI has dramatically lowered the barriers.

Today’s AI tools like Claude Code, Cursor, and GitHub Copilot can:

• Write functional code from a simple description

• Explain complex code in plain language

• Debug and fix errors

• Build entire applications with minimal guidance

Tasks that once required a team of five programmers might now be accomplished by two or three with AI assistance. More importantly, people who never thought they could create software are now building tools to solve their own problems.

The “no-code” and “low-code” movements are being supercharged by Generative AI, making software creation accessible to people without technical backgrounds.

Is Generative AI Just Another Tech Bubble?

Before investing your time in learning any technology, it’s smart to ask whether it has staying power. We evaluated Generative AI against five questions:

1. Does it solve real-world problems?
Yes. As we just saw, it’s already making a real difference in customer service, education, content creation, and software development. These aren’t trivial applications.

2. Is it useful in everyday life?
Yes. Unlike some technologies that only benefit specialized industries, Generative AI tools are immediately useful to almost everyone. Writing emails, learning new skills, creating content, understanding complex information. Daily value.

3. Is it creating economic impact?
Absolutely. Private AI investment hit $285 billion in the US alone in 2025, growing over 127% in a single year. Generative AI captures nearly half of all private AI funding globally. Major companies are restructuring entire divisions around AI capabilities.

4. Is it creating new job opportunities?
Yes. While there are legitimate concerns about job displacement, Generative AI is also creating entirely new career paths. The role of “AI Engineer” barely existed a few years ago. Now it’s one of the fastest-growing job categories.

New roles are emerging in prompt engineering, AI ethics, AI training and education, and specialized AI application development across industries.

5. Is it accessible to ordinary people?
Yes. This might be the most important part. Unlike previous waves of technology that required specialized knowledge, today’s Generative AI tools are designed to be used through natural language.

You don’t need to code or understand complex math. You simply talk to them in English, Hindi, Urdu, or whatever language you prefer. The technology is accessible to almost everyone, regardless of technical background.

All five answered yes. Generative AI is following the trajectory of truly transformative technologies like the internet, not temporary hype cycles.

Why This Matters Now

The Generative AI shift isn’t coming. It’s already here. But we’re still in the early days, comparable to where the internet was in the mid-90s. The most significant impacts and opportunities are still ahead.

By understanding these tools now, you position yourself to benefit from them rather than being caught off guard as they transform more aspects of work and life.

The good news? These tools are designed to be intuitive. You don’t need to understand everything about how they work to start benefiting from them today.

Coming Up

Now that you know what Generative AI is and why it matters, the natural next question is: what’s actually powering these tools? In our next article, we’ll break down Foundation Models, the massive pre-trained systems that make ChatGPT, Claude, and Gemini possible. If you’ve ever wondered why some AI tools are smarter than others, that one’s for you.

Inspired in part by CampusX’s Hindi-language AI education content.

AI for Common Folks — Making AI understandable, one concept at a time.

Subscribe now

Leave a comment

Good morning, someone threw a Molotov cocktail at Sam Altman’s San Francisco home and he responded with a blog post calling for de-escalation, the US government is now encouraging Wall Street banks to test Anthropic’s Mythos model while UK regulators scramble to assess its risks, and North Korean hackers compromised a widely used developer tool that exposed OpenAI’s macOS app signing certificates. Here’s what happened 👇

1. Someone Threw a Molotov Cocktail at Sam Altman’s Home

Someone allegedly threw a Molotov cocktail at OpenAI CEO Sam Altman’s San Francisco home early Friday morning. No one was hurt. Police later arrested a suspect at OpenAI headquarters, where he was threatening to burn down the building. The incident came days after a lengthy New Yorker profile by Ronan Farrow and Andrew Marantz that questioned Altman’s trustworthiness, with sources describing “a relentless will to power” and “a sociopathic lack of concern for the consequences” of deceiving people. Altman published a blog post Friday night, acknowledging mistakes and a tendency toward being “conflict-averse.” He called the New Yorker piece “incendiary” and said he had “underestimated the power of words and narratives.” He also invoked a Lord of the Rings metaphor, arguing that no one should try to “control AGI” and that the technology should be shared broadly.

Why it matters: The physical attack on a tech CEO’s home is a crossing of a line. AI anxiety is real and growing, and no matter what you think of Altman or OpenAI, Molotov cocktails are not criticism. But the deeper story is the New Yorker profile itself. More than 100 sources raised questions about whether the person steering the most ambitious AI company in the world can be trusted with that responsibility. That question is not going away.

Source: TechCrunch

2. US Pushes Banks to Test Mythos While UK Regulators Scramble

The Anthropic Mythos saga escalated on two fronts this weekend. Bloomberg reported that Treasury Secretary Bessent and Fed Chair Powell did not just warn bank executives about Mythos at their emergency meeting. They encouraged the banks to test it for defensive cybersecurity purposes. Goldman Sachs, Citigroup, Bank of America, and Morgan Stanley are now reportedly testing the model alongside JPMorgan Chase, which was one of the original 40 partner organizations. Separately, the Financial Times reported that UK financial regulators, including the Bank of England and the Financial Conduct Authority, are holding urgent talks with the National Cyber Security Centre to assess Mythos risks. British banks, insurers, and exchanges are expected to be briefed on those risks within two weeks.

Why it matters: The US government is now actively pushing banks to use a model from a company that the Pentagon blacklisted as a supply chain risk. That contradiction tells you everything about how fast AI cybersecurity is moving. Governments are not choosing between fear and adoption. They are doing both at the same time. The UK response shows this is not just an American problem. If a model can find vulnerabilities in every major operating system, every financial regulator on earth needs a plan for it.

Source: TechCrunch | Source: Reuters

3. North Korean Hackers Hit OpenAI Through Axios Supply Chain Attack

OpenAI disclosed Friday that a widely used developer library called Axios was compromised on March 31 as part of a broader software supply chain attack believed to be linked to North Korea. The attack caused a GitHub Actions workflow used by OpenAI to download and execute a malicious version of Axios. That workflow had access to a certificate and notarization material used for signing macOS applications, including ChatGPT Desktop, Codex, Codex-cli, and Atlas. OpenAI said its analysis concluded the signing certificate was likely not successfully stolen, and no user data was accessed. But as a precaution, the company is updating all security certifications and requiring macOS users to update their apps. Older versions of OpenAI’s macOS desktop apps will stop receiving updates or support after May 8 and may stop working entirely.

Why it matters: Supply chain attacks are the hardest kind of cybersecurity threat to defend against because you are not being attacked directly. You are being attacked through a tool you trust. Axios is one of the most widely used JavaScript libraries in the world. If North Korean hackers can compromise it, they can reach thousands of companies at once. OpenAI caught this one, but the broader lesson is that every company building AI is a target, and the tools they depend on are the weakest link.

Source: Reuters

Quick Hits

• Apple is testing four designs for AI-powered smart glasses it plans to sell in 2027, with a possible unveiling later this year, Bloomberg’s Mark Gurman reported. The glasses will not have displays but will support photos, video, phone calls, music, and Siri. Source: TechCrunch

• Claude dominated the conversation at the HumanX AI conference in San Francisco this week, with vendors and panelists repeatedly naming Anthropic’s chatbot as their tool of choice over ChatGPT. TechCrunch reported that the perception OpenAI has “fallen off” is becoming widespread among enterprise users. Source: TechCrunch

• South Africa unveiled a draft national AI policy seeking public comment on sweeping proposals to regulate and accelerate AI adoption, including new institutions and incentive programs. Source: Reuters

That’s it for today. The week ended with a physical attack on a tech CEO, two governments racing to figure out the same AI model, and a reminder that the software supply chain is only as strong as its weakest dependency. AI is no longer a technology story. It is a security story, a political story, and now, a personal safety story.

Forward this to someone who needs to stay in the loop.

Subscribe now

Good morning, the Treasury Secretary and Fed Chair held an emergency meeting with Wall Street’s top bankers over the cyber risks posed by Anthropic’s new Mythos model, Anthropic is quietly exploring whether to design its own AI chips as its revenue triples, and TSMC just posted record first-quarter revenue on the back of insatiable AI demand. Here’s what happened 👇

1. Treasury and Fed Warn Wall Street CEOs About Anthropic Mythos Risks

U.S. Treasury Secretary Scott Bessent and Federal Reserve Chair Jerome Powell convened an urgent meeting with bank CEOs this week to warn them about cybersecurity risks posed by Anthropic’s newly launched Mythos model. Anthropic has said the model is capable of identifying and exploiting weaknesses across “every major operating system and every major web browser.” Access to Mythos is limited to about 40 technology companies, including Microsoft and Google. CEOs from Citigroup, Morgan Stanley, Bank of America, Wells Fargo, and Goldman Sachs attended the Washington meeting. JPMorgan’s Jamie Dimon could not join. Anthropic proactively briefed senior U.S. government officials on the model’s capabilities before release.

Why it matters: This is the first time the Treasury Secretary and the Fed Chair have jointly called in bank CEOs over the capabilities of a single AI model. It signals that AI cybersecurity is no longer a tech problem. It is a financial stability problem. If a model can find zero-day vulnerabilities at scale, the institutions that hold the most money are the most obvious targets. The meeting was not about regulating Anthropic. It was about making sure the banks are ready for what comes next.

Source: Reuters

2. Anthropic Explores Designing Its Own AI Chips as Revenue Hits $30 Billion

Anthropic is in the early stages of exploring whether to design its own custom AI chips, Reuters reported, citing three sources. The plans are preliminary, no dedicated chip team has been assembled, and the company may ultimately decide to keep buying chips from others. Anthropic currently relies on Google’s TPUs, Amazon’s Trainium chips, and now CoreWeave’s Nvidia-based infrastructure. Its annualized revenue has surged to over $30 billion, up from roughly $9 billion at the end of 2025. Earlier this week, Anthropic signed a long-term compute deal with Google and Broadcom, and on Friday it added a new multi-year agreement with CoreWeave for cloud capacity to run its Claude models.

Why it matters: Designing a cutting-edge AI chip costs around half a billion dollars just to get started, so this is not a decision any company makes lightly. But the math is changing. When your revenue triples in four months and your biggest bottleneck is chip supply, building your own starts to look less like a vanity project and more like a survival strategy. Anthropic would join Meta and OpenAI in the custom chip race, and it would reduce its dependence on any single supplier at a time when every chip in the world has a waiting list.

Source: Reuters

3. TSMC Q1 Revenue Surges 35% to $35.7 Billion on AI Demand

TSMC, the world’s largest contract chipmaker, reported first-quarter revenue of T$1.134 trillion ($35.7 billion), a 35% jump year-over-year that beat market forecasts. The company said the growth was driven by unabated demand for AI applications. TSMC’s shares have gained 29% this year. Analysts expect the next quarter to be even stronger, with revenue forecasts rising to a record T$1.2 trillion. The results come as Middle East conflict is raising energy costs and could disrupt semiconductor supply chains, but demand for advanced AI chip production has so far overwhelmed those headwinds. TSMC’s customer Foxconn, Nvidia’s biggest server maker, also reported 30% first-quarter revenue growth.

Why it matters: TSMC makes the chips that make AI possible. When its revenue surges 35% and analysts expect the next quarter to be bigger, it tells you one thing: the companies building AI are not slowing down their spending. Every dollar that Meta, Amazon, Anthropic, and others commit to AI infrastructure eventually flows through TSMC’s fabs. The chip shortage is not easing. It is getting more expensive.

Source: Reuters

Quick Hits

• CoreWeave signed a multi-year cloud deal with Anthropic to supply compute capacity for Claude models. Financial terms were not disclosed. CoreWeave shares rose more than 5% in premarket trading, adding to a 29% gain this year. Source: Reuters

• The EU is set to classify ChatGPT as a “very large search engine” under the Digital Services Act, according to Germany’s Handelsblatt. The designation would subject OpenAI to stricter regulation in Europe, including content moderation and transparency requirements. Source: Reuters

• Elon Musk’s xAI sued Colorado to block the state’s new AI law from taking effect on June 30, arguing it violates the First Amendment and would force changes to Grok. The lawsuit intensifies the fight over whether AI regulation should be handled by states or Washington. Source: Reuters

• Canada’s Cohere and Germany’s Aleph Alpha are in merger talks, Handelsblatt reported, with Berlin backing the potential deal. If completed, it would create one of the largest non-U.S. AI companies. Source: Reuters

That’s it for today. The federal government called Wall Street into a room to talk about one AI model. Anthropic’s revenue tripled in four months and it still cannot get enough chips. TSMC’s record quarter proves that every company making these bets is making them bigger, not smaller. The AI arms race is now a financial infrastructure problem, and the people who regulate banks just admitted it.

Forward this to someone who needs to stay in the loop.

Subscribe now

Leave a comment

Good morning, Meta just signed a $21 billion compute deal that gives it early access to Nvidia’s next-generation chips, Amazon disclosed for the first time that its AI business is pulling in $15 billion a year, and Elon Musk dropped his $134 billion claim against OpenAI and offered to give everything to the nonprofit instead. Here’s what happened 👇

1. Meta Signs $21 Billion CoreWeave Deal, Gets Early Access to Nvidia’s Next-Gen Chips

Meta deepened its partnership with CoreWeave on Thursday with a fresh $21 billion cloud computing deal that extends through 2032. This comes on top of the $14.2 billion agreement the two companies signed in September. As part of the deal, Meta gets early access to Nvidia’s next-generation Vera Rubin chips, which are reportedly twice as fast as the current Blackwell platform. Meta plans to spend up to $135 billion on its AI buildout this year, the largest capital commitment of any tech company in history. CoreWeave, which went public last year and counts Microsoft as its biggest customer, said Meta is now among its largest clients. The deal landed one day after Meta unveiled Muse Spark, the first model from Meta Superintelligence Labs, an expensive team it assembled last year after the poor performance of its Llama 4 model.

Why it matters: The total value of Meta’s CoreWeave commitments is now north of $35 billion, locked in through the end of the decade. These are not research budgets. This is a company that makes its money from ads committing more capital to AI compute than most countries spend on their militaries. The Vera Rubin access is the strategic prize: it means Meta will be training on chips that competitors cannot buy yet. When a social media company is spending more on chips than Boeing spends on planes, the AI arms race is no longer a metaphor.

Source: Reuters

2. Amazon Reveals AWS AI Revenue Is $15 Billion a Year, Custom Chips Hit $20 Billion

Amazon CEO Andy Jassy disclosed for the first time that AWS AI services are generating annualized revenue of more than $15 billion, roughly 10 percent of AWS’s total $142 billion revenue run rate. In his annual shareholder letter, Jassy also revealed that Amazon’s custom chip business, including Graviton processors, Trainium AI chips, and Nitro networking cards, has doubled its annualized revenue to over $20 billion, up from $10 billion just one quarter earlier. Jassy hinted that demand for Amazon’s custom chips is so strong the company may start selling them directly to outside customers. “There’s so much demand for our chips that it’s quite possible we’ll sell racks of them to third parties in the future,” he wrote. Amazon is planning $200 billion in capital spending this year, mostly on AI.

Why it matters: This is the first time Amazon has put a dollar figure on its AI revenue, and it answers the question investors have been asking for two years: is this spending paying off? At $15 billion and growing, the answer is clearly yes. But the custom chip number is the more interesting signal. At $20 billion in annualized revenue, Amazon’s homegrown chip business is already larger than many standalone semiconductor companies. If Amazon starts selling those chips to third parties, it becomes a direct competitor to Nvidia in the data center, not just a customer.

Source: Reuters

3. Musk Drops $134 Billion Claim, Offers to Give All OpenAI Damages to the Nonprofit

Elon Musk amended his lawsuit against OpenAI and Sam Altman on Tuesday, dropping his demand for up to $134 billion in personal damages and instead asking the court to return all recovered funds to OpenAI’s original nonprofit charity. Musk’s lawyer said he is “not seeking a single dollar for himself.” The pivot came after US District Judge Yvonne Gonzalez Rogers denied Musk’s request for punitive damages and ruled that his expert’s calculations did not support his arguments for pocketing the money. Musk is still suing to unseat Altman from the board, unwind OpenAI’s for-profit conversion, and permanently restore the company to its original nonprofit structure. OpenAI called the filing “nothing more than a harassment campaign driven by ego, jealousy, and a desire to slow down a competitor.” The trial is expected to begin later this month.

Why it matters: This is not a change of heart. It is a legal strategy forced by a judge who was about to gut his case. Musk had been trying to pocket the damages himself, and the judge said no. Now he is reframing the entire suit as a public interest case, not a financial one. The question the jury will face is whether OpenAI’s pivot from nonprofit to for-profit was a breach of charitable trust. If Musk wins on that argument, it would not just cost Altman the company. It could set a precedent that constrains every future AI lab that starts as a nonprofit and converts.

Source: Ars Technica

Quick Hits

• Florida’s Attorney General launched an investigation into OpenAI and ChatGPT on Thursday. The state joins a growing list of regulators taking a closer look at the company’s practices ahead of its planned IPO. Source: Reuters

• OpenAI paused its UK data center project over concerns about regulation and costs, according to Reuters. The company had been planning a major buildout in Britain but is now reconsidering. Source: Reuters

• OpenAI projects $2.5 billion in advertising revenue this year and $100 billion by 2030, according to Axios. The company is betting that ads inside ChatGPT and its other products will become a major revenue stream alongside subscriptions. Source: Reuters

• Meta’s Superintelligence Lab unveiled its first public model, Muse Spark, with strong benchmarks but admitted “performance gaps” in agentic and coding tasks. The model is the first output from the costly team Meta built after Llama 4 underperformed. Source: Ars Technica

That’s it for today. Meta is writing $35 billion in checks to one cloud company. Amazon is revealing that its homegrown chips are a $20 billion business. And the trial that could decide whether OpenAI stays a for-profit company is about to begin. The money flowing into AI infrastructure is now so large it is reshaping the companies that spend it as much as the technology they are building.

Forward this to someone who needs to stay in the loop.

Subscribe now

Leave a comment

The Reality

A team member had an idea on Tuesday. By Thursday, it was a working product. She doesn’t know how to code. Not a single line.

That’s not a hypothetical. That’s what’s happening right now with what’s being called “vibe coding,” describing what you want to build in plain language and having AI write the actual code for you.

36% of new companies are now solo-founded. One person. No co-founder, no team. Five years ago, that number was 23%. The reason isn’t that founders got smarter. It’s that the tools caught up to their ambition.

Google’s research team has been working on something they call “generative UI.” You describe what you want, and in about a minute, you get a fully interactive application: buttons, logic, interface, all working. Yasi Matias, who leads Google Research, has watched this evolve firsthand. “People can now actually say what they want to be developed,” he said. “You can already develop an application that previously required a team.”

The Shift

The Old Way: Have an idea. Write a spec. Hire a developer. Wait weeks. Review. Iterate. Ship months later. Most ideas die in that gap between “I thought of something” and “someone built it.”

The New Reality: Describe what you want. Watch it get built. Test it. Ship it. The gap between idea and prototype collapsed from months to hours.

This changes who gets to build things. It’s not just developers anymore. The person who runs your social media, the operations lead who sees a workflow problem, the sales rep who wants a better tracking tool. They can all build now. The bottleneck shifted from “can you code?” to “do you have a clear enough vision of what you want?”

That’s a fundamentally different barrier. Technical skill was expensive and scarce. Clarity of vision is free and abundant, if you practice it.

The companies that understand this are going to look very different in two years. Instead of waiting for the engineering team’s backlog to clear, anyone with a problem and a clear description of the solution can ship a prototype the same week.

What To Do Next

If you have an idea you’ve been sitting on, whether it’s a tool, a workflow, a product, stop waiting for someone to build it. Open Claude, ChatGPT, or Cursor and describe what you want. Start with something small. A dashboard. A calculator. A form that automates something you do manually.

You’ll be surprised how far you can get without writing a single line of code. The worst case is you learn what’s possible. The best case is you ship something real.

And if you’re a manager, start asking your team: “What would you build if you could?” The answers might be worth more than your next engineering hire.

The One Thing to Remember

The barrier between “I have an idea” and “I built it” is gone. The new competitive advantage isn’t coding ability. It’s the clarity to describe exactly what you want and the judgment to know if what you got back is good enough.

This insight comes from “Google VP: The AI Shift Is Done and the Gap Between People Is Growing” featuring Yasi Matias, head of Google Research. The AI Shift curates wisdom from AI leaders for busy professionals navigating the AI era. What’s one thing you’d build if the technical barrier didn’t exist?

Subscribe now

Leave a comment

Good morning, Anthropic dropped a new frontier model into the hands of 12 companies to hunt zero-day vulnerabilities, Intel signed on to Elon Musk’s most ambitious chip project yet, and a fresh test of Google’s AI Overviews puts the error rate at 10 percent. Here’s what happened 👇

1. Anthropic Quietly Hands “Mythos” to Microsoft, Apple, and Amazon for Cybersecurity Work

Anthropic on Tuesday released a preview of a new frontier model called Mythos as part of a security initiative it is calling Project Glasswing. Twelve partner organizations are getting first access: Amazon, Apple, Broadcom, Cisco, CrowdStrike, the Linux Foundation, Microsoft, and Palo Alto Networks among them. The model is being used to scan first-party and open source software for code vulnerabilities, and Anthropic claims that in just the past few weeks, Mythos has already identified “thousands of zero-day vulnerabilities, many of them critical,” with some of the bugs sitting undiscovered in code for one to two decades. A previously leaked internal memo described Mythos as “one of the most powerful” models the company has ever built, and Anthropic says it has been in “ongoing discussions” with federal officials about deploying it, though those talks are complicated by the company’s active legal fight with the Trump administration.

Why it matters: This is the first real glimpse of what frontier AI looks like when you point it at the security of the software the world actually runs on. Decades-old bugs that humans missed are now being surfaced in weeks. That cuts both ways. The same capability that helps Microsoft patch Windows also helps an attacker find the same hole first. Anthropic chose to give it to defenders, but the gap between defensive and offensive use of these models is shrinking by the month.

Source: TechCrunch

2. Intel Joins Elon Musk’s Terafab to Build the Chips Powering Humanoid Robots

Intel announced Tuesday that it is joining Terafab, Elon Musk’s chip-manufacturing megaproject with SpaceX and Tesla, with the stated goal of producing one terawatt of compute per year for AI and robotics. The handshake came after Intel CEO Lip-Bu Tan hosted Musk at Intel’s campus over the weekend. Musk has previously laid out plans to build two advanced chip factories in Austin, Texas: one to power Tesla cars and humanoid robots, the other to feed AI data centers in space. Intel’s stock jumped more than 2 percent on the news. For Intel, which lost $10.32 billion in its foundry business last year, the deal is a lifeline for its turnaround story and a chance to prove its 18A manufacturing tech can win the largest customers.

Why it matters: A terawatt per year is a number that did not exist in the chip industry before this week. To put that in scale, the entire global semiconductor industry today produces a small fraction of that. Musk is betting that humanoid robots and orbital data centers will need so much silicon that the only way to get there is to build the factories himself. Intel just bet its turnaround on that future being real.

Source: Reuters

3. Google AI Overviews Tell “Millions of Lies Per Hour,” New Study Finds

A New York Times analysis published Tuesday tested Google’s AI Overviews using SimpleQA, a benchmark with more than 4,000 verifiable factual questions, and found the system gets roughly 1 in 10 answers wrong. Extrapolated across all Google searches, that works out to tens of millions of incorrect answers per day. The study, run with help from AI startup Oumi, showed accuracy improved from 85 percent under Gemini 2.5 to 91 percent after the Gemini 3 update, but the misses are striking. Asked when Bob Marley’s home became a museum, AI Overviews picked the wrong year from a Wikipedia page that listed two. Asked when Yo Yo Ma was inducted into the Classical Music Hall of Fame, it cited the organization’s own website and then claimed the hall does not exist. Google pushed back, saying the SimpleQA test “has serious holes” and does not reflect what people actually search for.

Why it matters: Nine out of ten sounds great until you realize Google handles billions of searches a day. AI Overviews now sits at the very top of the results page, ahead of the blue links it cites, which means most users never check the source. The product is designed to make you stop reading right there. When the answer is wrong, that confidence becomes a problem. And the trade-off Google made is buried in the model selection: the fast, cheap Gemini Flash model handles most queries, not the more accurate Pro model, because speed wins on a search page.

Source: Ars Technica

Quick Hits

• Anthropic also expanded its compute deal with Google and Broadcom this week, locking in 3.5 gigawatts of new TPU capacity coming online in 2027. The expansion is part of Anthropic’s $50 billion U.S. infrastructure commitment, and comes as the company’s run-rate revenue surges past $30 billion. Source: TechCrunch

• Uber became the latest major company to switch to Amazon’s custom AI chips, using AWS Trainium2 to train the AI models that power its ride and delivery business. Another sign that Nvidia’s grip on AI training is loosening at the top of the market. Source: Reuters

• PIMCO is weighing a $14 billion debt deal to finance Oracle’s new Michigan data center, according to Bloomberg. AI infrastructure is now being financed at scales that used to belong to oil pipelines and toll roads. Source: Reuters

• Atlassian launched visual AI tools and third-party agents inside Confluence, letting AI agents from outside vendors operate directly inside the workspace where teams already write docs and run projects. Source: TechCrunch

That’s it for today. Yesterday OpenAI was pitching the public on robot taxes and a four-day workweek. Today Anthropic is quietly handing decade-old security bugs to Microsoft, Intel is signing onto a one-terawatt chip project, and Google is being told its flagship AI is wrong tens of millions of times a day. The public-facing story and the actual buildout are running on different tracks, and the buildout is moving faster.

Forward this to someone who needs to stay in the loop.

Subscribe now

Leave a comment

AI won’t make everyone equal. It will make the gap between exceptional and average wider than ever.

The Reality

There’s a comforting story floating around the tech world right now. It goes like this: AI tools will level the playing field. Junior developers will code like seniors. Non-technical people will build like engineers. Everyone will be elevated.

Max Brodeur-Urbas, founder of Gumloop, an automation platform processing 4 million workflows daily for Instacart, Shopify, and DoorDash, sees something different happening.

“It’s possible that the last generation of great engineers has been born,” he says. “Because there was this era of actually needing to understand what’s going on and then getting accelerated by AI. But now people can skip the understanding part and just accelerate.”

The people who learned the fundamentals first, who understand why things work and not just how to make them work, are now getting turbocharged by AI. They’re becoming exceptional at a speed that wasn’t possible before.

Everyone else is generating slop.

The Shift

AI creates a fork in the road, not a rising tide.

On one path: people who use AI as a learning tool. They pause. They try to understand the problem. They ask AI to teach them what they don’t know. They build on genuine comprehension.

On the other path: people who use AI as a shortcut. Their website works. The feature does what they wanted. But they never took the time to understand why it worked, what could break, or what knock-on effects it might create.

The Old Way: Everyone needed to learn the fundamentals. The bar was high. Progress was slow but solid.
The New Reality: You can skip the fundamentals entirely. Your code compiles. Your app runs. But you’re building on a foundation you can’t see, can’t debug, and can’t improve.

“It’s so easy to just not want to understand why something works,” Max says. “Your website worked, the feature did what you wanted, but you didn’t take the time to really dig into why.”

This is the trap. AI makes it effortless to skip understanding. And skipping understanding feels productive in the moment. You shipped the feature. You launched the product. But when something breaks in a way the AI can’t fix, you’re stuck.

“If you can actually have the determination to pause, try to understand the problem, have AI teach you the things you don’t understand, you’ll become exceptional even faster than before. And then the average person will just kind of fall to the slop.”

What To Do Next

The next time AI gives you a solution that works, don’t move on. Spend five minutes understanding why it works.

Ask the AI to explain its reasoning. Change one variable and see what breaks. Read the output instead of just running it.

This is the new competitive advantage. Not using AI. Everyone will use AI. The advantage is using AI while maintaining the discipline to actually learn from it.

The engineers, marketers, analysts, and operators who do this will pull ahead so fast that the gap becomes permanent. The ones who don’t will produce work that looks right on the surface and falls apart under pressure.

The One Thing to Remember

AI doesn’t level the playing field. It amplifies the gap. The people who understand the fundamentals and use AI to go faster will become unreachable. The people who skip understanding will produce impressive-looking work that breaks at the first unexpected input.

This insight comes from “50 AI Agents Running My Company Is a Lie” featuring Max Brodeur-Urbas, founder of Gumloop. The AI Shift curates wisdom from AI leaders for busy professionals navigating the AI era. Are you using AI to learn faster, or to skip learning entirely?

Subscribe now

Leave a comment

Good morning, OpenAI dropped a utopian policy blueprint for the AI economy, The New Yorker dropped a 100-source investigation questioning whether Sam Altman can be trusted to deliver any of it, and Iran posted satellite imagery of the Stargate data center with a threat attached. Here’s what happened 👇

1. OpenAI’s Wish List: Robot Taxes, Public Wealth Funds, and a Four-Day Workweek

OpenAI released a 30-page policy document titled “Industrial Policy for the Intelligence Age” laying out how it thinks governments should handle the economic fallout of superintelligent AI. The proposals are striking because they read more like a Bernie Sanders white paper than a Silicon Valley wish list. OpenAI suggests shifting the tax burden from labor to capital, floating a “robot tax” that would force AI systems to pay the same payroll taxes as the human workers they replace. It proposes a Public Wealth Fund that would give every American an automatic stake in AI companies, with returns distributed directly to citizens. It calls for subsidized 32-hour, four-day workweek pilots with no loss in pay, expanded retirement matches, employer-covered childcare, and portable benefits that follow workers across jobs. The document acknowledges that AI-driven growth could “hollow out the tax base that funds Social Security, Medicaid, SNAP, and housing assistance” if nothing changes.

Why it matters: This is the $852 billion company that built ChatGPT openly admitting that the current economic model cannot survive the technology it is selling. When the people building the thing tell you it will gut the tax base, replace the workers, and require a robot tax to fix it, that is not a marketing pitch. That is a confession dressed up as policy. The question is whether anyone in Washington is going to take a redistribution agenda seriously when it comes from a for-profit company whose CEO has spent the last year lobbying against AI safety laws.

Source: TechCrunch

2. “The Problem Is Sam Altman”: New Yorker Investigation Lands the Same Day

Hours after OpenAI published its policy vision, The New Yorker published a massive investigation into whether Sam Altman can be trusted to deliver on any of it. The reporters interviewed more than 100 people familiar with how Altman operates, reviewed internal memos, and interviewed Altman himself more than 12 times. The portrait is brutal. One board member described Altman as having “two traits that are almost never seen in the same person. The first is a strong desire to please people, to be liked in any given interaction. The second is almost a sociopathic lack of concern for the consequences that may come from deceiving someone.” Internal messages from former chief scientist Ilya Sutskever and former research head Dario Amodei (now CEO of Anthropic) document what they called “an accumulation of alleged deceptions and manipulations.” Amodei wrote bluntly: “The problem with OpenAI is Sam himself.” One current OpenAI researcher told The New Yorker that Altman “sets up structures that, on paper, constrain him in the future. But then, when the future comes and it comes time to be constrained, he does away with whatever the structure was.”

Why it matters: The timing is not a coincidence. OpenAI’s chief global affairs officer told The Wall Street Journal that the company is urgently concerned about negative public opinion. The policy document reads like an attempt to reset a narrative that is slipping. But trust is the entire product when you are asking the public to let you build superintelligence. If the people who worked closest with Altman are saying out loud that he tells everyone what they want to hear and then walks away from the constraints he agreed to, no policy white paper fixes that.

Source: Ars Technica

3. Iran Threatens to Bomb the $500B Stargate Data Center in Abu Dhabi

Iran’s military released a video this weekend showing satellite imagery of OpenAI’s Stargate data center in the United Arab Emirates, with a message that read “nothing stays hidden to our sight, though hidden by Google.” Military spokesperson Ebrahim Zolfaghari warned that if the U.S. follows through on threats to strike Iranian power and water infrastructure, Iran will hit U.S. tech and energy infrastructure across the Middle East in return. This is not an empty threat. Iranian missiles have already struck AWS data centers in Bahrain and an Oracle data center in Dubai earlier in the war that began in February. Iran has also publicly named Nvidia and Apple as targets. Stargate is the $500 billion joint venture between OpenAI, SoftBank, and Oracle to build out global AI infrastructure, originally announced in January 2025. The Trump administration has threatened further strikes on Iranian civilian infrastructure if Iran does not reopen the Strait of Hormuz by Tuesday.

Why it matters: AI data centers are no longer just real estate. They are now strategic military targets, the way oil refineries became targets in the 20th century. The race to build AI infrastructure overseas, especially in the Gulf, was supposed to solve power and land constraints at home. Instead it has put the most expensive computing assets in the world in the middle of an active war zone. Every company building toward “agentic AI” depends on physical buildings that can be hit by a missile. That is the part of the AI boom no one prices in.

Source: TechCrunch

Quick Hits

• Samsung said Q1 operating profit will jump roughly eightfold on red-hot AI chip prices, a quarterly record that nearly equals what the company earned in all of last year. Source: Reuters

• Robotics company Generalist released GEN-1, a new physical AI model hitting 99% success rates on tasks like folding boxes, packing phones, and servicing robot vacuums. The model can improvise when objects move unexpectedly: “Nobody has programmed the robot to make mistakes, therefore nobody has programmed the robot to recover from mistakes. And that just happens for free.” Source: Ars Technica

• OpenAI is asking the California and Delaware attorneys general to investigate Elon Musk for what it calls “anti-competitive behavior” related to xAI and his ongoing legal battles with the company. Source: Reuters

• Google quietly launched an offline-first AI dictation app on iOS, processing speech-to-text on-device with no cloud roundtrip. A small but meaningful shift toward local AI on phones. Source: TechCrunch

That’s it for today. OpenAI wants you to imagine a future where AI funds your retirement and gives you a four-day workweek. The same week, the people who built the company are telling reporters they don’t trust the man pitching it, and the buildings that would deliver it are being targeted by missiles. The vision and the reality are diverging fast.

Forward this to someone who needs to stay in the loop.

Subscribe now

Leave a comment