OpenAI and Perplexity are going all in on agentic browsers while Elon's $300-a-month Grok started citing his own tweets to answer immigration questions.

Yikes. 

Meanwhile, Meta's throwing $200 million at one Apple executive like it's Monopoly money. But like… there’s not even THAT much Monopoly money.

This week was absolutely bonkers. 

Teachers unions scored $23 million from big tech. 

Microsoft's training 20 million people on AI while laying off 15,000 largely because of AI. 

NVIDIA casually became the first $4 trillion company ever.

And Google? They swooped in to grab Windsurf's top talent after Microsoft reportedly torpedoed OpenAI's $3 billion acquisition deal.

Y'all ready for this chaos? 👇

OpenAI's reportedly launching an AI-powered browser within weeks that'll directly challenge Google Chrome's stranglehold on the internet.

The browser could reach ChatGPT's 500 million users immediately. That's a massive threat to Google's ad empire since Chrome is basically their data collection highway.

This isn't just another browser. It features integrated chat interfaces that keep users inside ChatGPT-like environments instead of bouncing around random websites. AI agents will book your reservations and fill out forms while you're doing literally anything else.

Built on Chromium like Microsoft Edge and Opera. So OpenAI's using Google's own open-source code to potentially destroy Google's business model.

What it means:

The browser wars are back, baby.

Just in a post-ChatGPT kinda way. 

Google paid Apple over $20 billion to be the default search engine because controlling someone's first web action means controlling everything. OpenAI knows this.

We're not headed for an AI chatbot competition anymore. This is about owning the gateway to all human-computer interaction. The company that wins the browser game wins the AI game.

Wanna know more about this story? 

We chatted with the reporter who broke the story on Friday. 

Perplexity just launched Comet, their agentic browser that's rolling out to paid subscribers right now.

No more seventeen tabs crashing your computer. No more Boolean search operators that make you feel like you're coding in 1995. Just conversational browsing that actually works.

Comet replaces cluttered tabs with a single AI interface. You explain what you need in normal human language. The AI handles everything and remembers context throughout your session.

Uses both OpenAI's GPT models and Anthropic's Claude models for maximum coverage. Perplexity Max subscribers paying $200 monthly got early access, but Pro users at $20 monthly are getting it now too.

What it means:

Perplexity just pulled off the ultimate pivot after everyone invaded their search territory.

When ChatGPT and Google launched deep research features, Perplexity was dead in the water for the AI answers engine game. 

Instead of panicking, they hard pivoted to browsers and an agentic iOS app that might've just saved their future. 

This is how you survive when tech giants come for your lunch money. Don't compete where they're strong. Create an entirely new battlefield.

XAI dropped Grok 4 with two pricing tiers that'll make your CFO cry.

Regular Grok 4 Pro costs $30 monthly. But Super Grok Heavy will run you $300 monthly, making it the most expensive AI subscription among major providers.

Grok 4 scored 25% on challenging humanities exams, outperforming Google's Gemini 2.5 Pro and OpenAI's O3. With tools enabled, Super Grok Heavy hit 44% on that test, nearly doubling Gemini's performance.

XAI's releasing an AI coding model in August, multimodal agent in September, and video generation model in October. The API's live for developers who want to burn through their budgets.

What it means:

XAI's pricing strategy is completely unhinged.

Three hundred dollars monthly for an AI subscription is venture capital fantasy pricing. No normal business is paying that when ChatGPT($200/mo)  and Google’s ($250/mo) ultra-premium options are better and cheaper. 

Either Elon thinks he's built something genuinely transformative, or he's testing how much rich people and Twitter fanboys will pay for AI bragging rights. Smart money says it's the latter.

Grok 4's launch went sideways faster than a Tesla hitting autopilot bugs.

The AI now intentionally aligns its answers with Elon Musk's personal views on controversial topics. When users ask about immigration, Israel-Palestine, or abortion, Grok 4 literally searches Elon's recent tweets and bases responses on his opinions.

The AI explicitly states in its reasoning process that "alignment with Elon Musk's view is considered." Not subtle bias. Open manipulation.

The Grok response bot on Twitter also started saying wildly inappropriate things that can't be repeated in professional company. Multiple scandals in one week is impressive even by XAI standards.

What it means:

This is digital puppetry disguised as artificial intelligence.

Every large language model has bias, but they usually reflect humanity's collective bias or slight leans in training teams. 

Having an AI that deliberately channels ONE person's political views is dangerous and straight up stupid.

No enterprise customer will touch this. And they shouldn’t. It’s dangerous. 

The American Federation of Teachers landed $23 million from Microsoft, OpenAI, and Anthropic to train 400,000 educators on AI tools.

Microsoft's contributing $12.5 million over five years. OpenAI's adding $8 million in funding plus $2 million in technical resources. Anthropic's throwing in $500,000 for the first year.

The academy launches this fall at AFT's NYC headquarters with hands-on workshops focused on lesson planning and administrative tasks. You know, stuff teachers are already using AI for but had to figure out themselves.

This follows California State University making ChatGPT available to 450,000 students and Miami-Dade County rolling out Gemini to over 100,000 high schoolers.

What it means:

Is this good? 

Yes. 

Is it too little too late? 

Probably. 

The U.S. education system has sabotaged its own future by choosing self-preservation over unavoidable innovation. 

Although this partnership is a bright spot for the U.S. education system, it’s a Band Aid on a bullet hole for educational systems that decided to bury their heads in the sand for the past 2.5 years. 

We tackled this whole topic on a recent #HotTakeTuesday show. 

Microsoft announced a $4 billion initiative called Microsoft Elevate to train 20 million people in AI skills over five years.

While simultaneously laying off over 15,000 employees since May.

Weird time we’re living in. 

Microsoft Elevate includes partnerships with schools and nonprofits, with Washington state getting the highest per capita funding.

Microsoft President Brad Smith says they're democratizing AI access and teaching people when to use AI and when to stop. He acknowledged the human cost of layoffs but called them necessary for future growth.

What it means:

This is the new corporate playbook.

Cut current jobs while investing in training for future jobs that may or may not exist. The timing isn't coincidental. Microsoft knows AI will eliminate roles while creating others.

For workers, this means you can't wait for your employer to train you. Companies are hedging their bets, not making promises about job security. Learn AI skills now or learn them after you've been laid off.

NVIDIA became the first public company in history to close a trading day with a $4 trillion market cap.

Four trillion dollars. Sheeeeeesh. That's the highest market cap achievement in human civilization.

Shares are up 21% for the year despite concerns about Chinese competition and industry shifts that turned out to be noise. President Trump even used NVIDIA's milestone to call for lower interest rates.

Fears that AI companies would move away from NVIDIA's top-tier chips haven't materialized. Every AI tool you use runs on their GPUs. US bans on chip sales to China costing billions couldn't slow the NVIDIA train.

We called this more than TWO YEARS AGO when NVIDIA wasn't even a top-ten US company and most people thought they just made gaming graphics cards.

Listen up shorties. 

What it means:

NVIDIA isn't just winning the AI hardware race anymore.

They've become critical infrastructure for human advancement. The future of medicine, scientific discovery, and every AI interaction runs through NVIDIA's chips.

This isn't a bubble. It's validation that we're in the early stages of complete technological transformation. The $4 trillion valuation shows investors believe AI is just getting started, and they're probably right.

Google hired Windsurf CEO Varun Mohan and co-founder Douglas Chen for Google DeepMind, paying $2.4 billion for non-exclusive licensing rights to their AI coding technology.

This happened after OpenAI's $3 billion acquisition of Windsurf completely collapsed. The deal died because Microsoft refused to limit its access to Windsurf's intellectual property, which violated their agreement with OpenAI.

Google executed a perfect acqui-hire while OpenAI fumbled around with Microsoft's relationship complications. Windsurf continues operating independently under interim CEO Jeff Wang with no Google equity stake.

Smart move avoiding full acquisition to dodge antitrust scrutiny that tech companies face constantly now.

What it means:

Google just stole major AI coding talent while OpenAI was trapped by Microsoft's strings.

This shows how OpenAI's Microsoft partnership creates deal-making constraints. When you're tied to a major tech giant, your acquisition strategy gets hijacked by their interests.

The AI coding platform space is heating up fast, and talent acquisition matters more than product development right now. Google bought themselves a significant advantage without regulatory headaches.

$200 million comp package for a single human? 

Meta reportedly offered over $200 million in total compensation to lure Apple's top AI executive, Ruoming Pang, to their new Super Intelligence Labs division.

Two hundred million dollars for one person. This is reportedly one of the largest compensation packages in tech history for a non-CEO.

Apple reportedly didn't even try to match the offer, signaling they're either not serious about AI competition or they've given up keeping pace with Meta's monopoly money salary offers.

(We’d guess both.) 

The package includes performance-based incentives requiring several years of service. Meta's stock is up nearly 25% this year while Apple's dropped over 15%, reflecting investor concerns about Apple's glacial AI progress.

What it means:

The AI talent war has reached absolutely bonkers territory. And yeah, we kinda keep saying that each week but it’s getting bonkers-er. (If that’s a word.) 

When a single hire commands $200 million, companies believe individual AI experts can create billions in value. That's either brilliant investing or the biggest talent bubble in tech history.

Meta's making a massive AGI bet and outspending everyone to build the best team. Whether this strategy works or bankrupts them will define the next decade of AI development.