👟 Allbirds sells shoes, buys GPUs

👟 Allbirds Ditches Shoes for GPUs, Stock Goes Brrrrr

Former sustainable shoe brand Allbirds announced a $50M deal to become a GPU-as-a-Service company called "NewBird AI." Shares rocketed from $3 to over $20 - a 600% surge. Shareholders vote next month to strip the company's "public benefit" status entirely.
Why it matters: When a shoe company can 6x its stock by slapping "AI" on the business model, the market isn't just frothy - it's fully delusional. Peak AI bubble energy right here.

🧮 GPT-5.4 Pro Just Solved Math That Stumped Humans for 60 Years

OpenAI's GPT-5.4 Pro produced a valid proof for a math problem that had been unsolved for nearly a century. Mathematician Jared Lichtman confirmed the AI found a path humans had completely overlooked.
Why it matters: This isn't "AI writes your essay" territory anymore - it's genuine scientific discovery. If LLMs can find proofs humans miss, they're not just tools. They're research collaborators. That's... actually huge??

🔥 Snap Fires 1,000 People, Wall Street Throws a Party

Snap cut 16% of its workforce (~1,000 people), with CEO Evan Spiegel saying AI now writes 65% of new code and handles over 1M monthly queries internally. The company is targeting $500M in annual savings by end of 2026. The stock rose 7-9%.
Why it matters: The pattern is now undeniable - announce AI-driven layoffs, stock goes up. Over 70,000 tech jobs have been cut this year. Wall Street loves efficiency. Workers... less so.

📰 Peter Thiel Wants a Jury of LLMs to Judge Journalists

A new startup called Objection, backed by Peter Thiel and Balaji Srinivasan, lets anyone pay $2,000 to challenge a news story using AI fact-checking. A "jury" of LLMs from OpenAI, Anthropic, xAI, Mistral, and Google evaluates the claim. A companion tool called "Fire Blanket" flags disputed stories on X in real time.
Why it matters: So now billionaires can pay to have AI score journalists' credibility?? Critics warn it could chill whistleblowing and intimidate reporters. Whether this is "accountability" or "weaponized AI for the powerful" depends entirely on who's writing the $2,000 check.

🇪🇺 EU Tells Meta Its WhatsApp AI Fee Is Illegal

The European Commission told Meta that charging third-party AI assistants a fee to operate on WhatsApp appears to violate EU antitrust rules. Meta's defense?? A "small bakery in France" shouldn't have to subsidize OpenAI's access. Italy's competition watchdog has also joined the investigation.
Why it matters: This is a major test case for AI gatekeeping on dominant platforms. If the EU forces open access, it could set the rules for how AI assistants compete across messaging apps worldwide. Meta's bakery analogy is... creative.

🏦 Trump Says There "Should Be" an AI Kill Switch for Banking

In a Fox Business interview, President Trump said AI could make banking "better and safer" but acknowledged the risks - and when asked about a government "kill switch" for AI in finance, he said "there should be." The comments followed warnings that Anthropic's new Mythos model could supercharge cyberattacks on banks with legacy systems. Anthropic says Mythos Preview won't be made generally available.
Why it matters: A rare moment of caution from a president who's otherwise been full-speed-ahead on deregulation. When the White House says AI in banking needs guardrails, that's... not nothing.

💰 Jane Street Drops $6 Billion on CoreWeave

Trading giant Jane Street committed ~$6 billion for CoreWeave's cloud services and made a $1 billion equity investment at $109/share (a 7% discount). This is CoreWeave's third massive deal in a week, after contracts with Anthropic and an expanded $21 billion deal with Meta. CoreWeave's market cap has hit $61.6 billion - up from $23 billion at IPO.
Why it matters: The demand for AI compute is absolutely insatiable. CoreWeave went from niche Nvidia-backed cloud provider to critical AI infrastructure in like... a year. The money printer is not slowing down.

📞 Someone Deepfaked Marco Rubio Over Signal

An unknown person used AI voice cloning to impersonate U.S. Secretary of State Marco Rubio, contacting three foreign ministers and two other U.S. officials via Signal to extract sensitive information.
Why it matters: Deepfake scams just went from "tricking grandma" to "infiltrating global diplomacy over encrypted messaging." This is lowkey terrifying.

💸 Anthropic Switches Claude Enterprise to Pay-Per-Token

Anthropic is moving Claude Enterprise from flat pricing to a token consumption model - meaning heavy users could see their bills jump significantly.
Why it matters: The "unlimited AI" era was fun while it lasted. Usage-based pricing is becoming the industry standard, and power users are about to find out exactly how expensive their Claude habit really is.

🤖 Cadence and Nvidia Team Up to Train Robots in Simulation

Cadence Design Systems and Nvidia are integrating Cadence's physics engines with Nvidia's AI models to train robots inside computer simulations. Cadence also launched an AI agent that handles physical chip layout design, now available on Google Cloud. Shares jumped 4%.
Why it matters: Training robots in the real world is slow, expensive, and occasionally destructive. Simulation with accurate physics data is how you get robots that can actually do useful things without breaking everything first.

📊 LinkedIn Says AI Isn't Killing Jobs Yet - But Your Job Is Changing

LinkedIn's data shows hiring is down ~20% since 2022, but the company says rising interest rates - not AI - are the main cause. They haven't seen outsized losses in AI-vulnerable fields like customer support. However, the skills needed for the average job have changed 25% in recent years, and LinkedIn expects that to hit 70% by 2030.
Why it matters: AI isn't replacing your job yet. It's replacing what your job is. The real disruption isn't a pink slip - it's your entire role description changing underneath you while you're still sitting in it.

🎨 Adobe Launches an AI Assistant That Actually Uses Photoshop

Adobe released the Firefly AI assistant, which can autonomously tap into Photoshop, Illustrator, and Premiere Pro to execute creative tasks from natural language instructions. It also connects to Anthropic's Claude.
Why it matters: Adobe's bet is that professionals still want premium, legally safe AI tools - not cheap slop generators. Whether that bet pays off depends on how long "legally safe" stays a selling point.

💼 Big Tech Dealmaking Hits a Five-Year High

Global M&A hit $1.2 trillion across 11,000+ deals in Q1 2026. OpenAI and Anthropic pulled in $110 billion and $30 billion in funding respectively. SpaceX is eyeing a $1.75 trillion valuation. Meanwhile, Anthropic is lobbying the Trump administration through Ballard Partners - the firm of White House Chief of Staff Susie Wiles - while simultaneously suing it.
Why it matters: The Trump era is open season for AI megadeals. And apparently you can sue the government and lobby it at the same time?? Noted.

🗣️ Google Drops Gemini 3.1 Flash TTS - Actually Good AI Speech

Google released Gemini 3.1 Flash TTS, a text-to-speech model with audio tags that let you control tone, pace, accent, and delivery using natural language. It supports 70+ languages, ranks #2 on Artificial Analysis's TTS leaderboard, and all output is watermarked with SynthID.
Why it matters: Expressive, controllable AI speech at scale is a game-changer for voice apps and podcasts. Also maybe don't combine this with the Rubio deepfake story... just saying.

🖥️ Google Launches a Gemini Desktop App (A Year Late)

Google released a native Mac app for Gemini with screen-sharing, Drive/Photos access, and image/video generation. A Windows version also launched. But the app still trails ChatGPT and Claude in agentic abilities - it's mostly a chat-first assistant while rivals can execute tasks directly on your machine.
Why it matters: The desktop AI assistant war is heating up and Google showed up a year late with a chat window. But Google's distribution advantage is massive - never count out the company that owns your browser, email, and search bar.

📱 India's Emergent Launches an AI Agent Inside WhatsApp

Indian startup Emergent (8M+ builders, $70M raised at $300M valuation) launched Wingman, an autonomous AI agent that runs inside WhatsApp, Telegram, and iMessage. It handles routine tasks across email, calendars, and workplace tools while asking for human approval on bigger decisions.
Why it matters: The best AI interface might not be a new app - it's the chat window you already have open. Smart play for global adoption, especially in markets where WhatsApp is the internet.

📚 AI Study App Gizmo Hits 13M Users

Gizmo, an AI-powered study platform that turns notes into quizzes and flashcards, grew from 300K to 13 million users across 120+ countries. It just raised a $22M Series A led by Shine Capital and plans to scale from 7 to ~30 employees while expanding in the U.S. college market.
Why it matters: With U.S. academic performance at historic lows, Gizmo is proving that gamified AI learning can actually compete with TikTok for students' attention. That's... genuinely impressive??

🔒 Gitar Raises $9M to Fix AI's Terrible Code Problem

Gitar, founded by an Intel/Google/Uber veteran, launched from stealth with $9M to tackle the flood of buggy AI-generated code. The platform deploys AI agents for code reviews, testing, and CI workflows, with the goal of making human code reviews the exception, not the rule.
Why it matters: AI is writing code faster than ever but the quality is... not great. Gitar's betting the next big opportunity isn't generating more code - it's making sure the code that ships doesn't catch fire in production.