In partnership with

Introducing the first AI-native CRM

Connect your email, and you’ll instantly get a CRM with enriched customer insights and a platform that grows with your business.

With AI at the core, Attio lets you:

• Prospect and route leads with research agents

• Get real-time insights during customer calls

• Build powerful automations for your complex workflows

Join industry leaders like Granola, Taskrabbit, Flatfile and more.

👉Try Attio Pro for free

AI's Latest Plot Twist: OpenAI Just Became Your Internet Middleman

Listen, I know what you're thinking. "Another week, another AI announcement." But here's the thing—OpenAI didn't just launch another chatbot feature last week. They launched a goddamn browser.

On October 21st, OpenAI unveiled ChatGPT Atlas, a Chromium-based web browser that puts AI directly between you and the internet. And before you dismiss this as just another product launch, let me tell you why this matters: whoever controls how you access the internet controls everything.

TLDR: OpenAI is making a power play for internet infrastructure, tech giants are burning cash on AI chips like there's no tomorrow, and we're all watching the moment when AI stops being a tool and starts being the entire interface.

The Real Estate Play You Didn't See Coming

Remember when we talked about how McDonald's is secretly a real estate company that happens to sell burgers? Well, OpenAI just pulled the same move—except instead of owning the land under the restaurant, they're trying to own the way you get to the restaurant.

ChatGPT Atlas isn't just a browser with some AI features slapped on. It's got:

• An "Ask ChatGPT sidebar" that's always there, watching, helping, remembering

• Browser memories that recall every page you've visited (cool, cool, not creepy at all)

• An agent mode for premium subscribers that can autonomously book your hotels, create documents, and basically act as your digital assistant

Available first on macOS with Windows, iOS, and Android coming soon, Atlas is OpenAI's direct shot at Google Chrome's dominance. And here's the kicker: it positions OpenAI to capture advertising revenue.

Think about that for a second. OpenAI has been burning through billions building the most sophisticated AI on the planet. Now they're positioning themselves to become the new Google—the middleman who gets a cut of every transaction, every ad impression, every search query. It's not about making a better browser. It's about controlling the tollbooth.

The Chip Wars: When "Big" Doesn't Even Cover It

While OpenAI was launching their browser, they were also casually announcing a $10 billion partnership with Broadcom to design proprietary AI processors.

Let's break down what $10 billion in custom chips actually means: We're talking about 10 gigawatts of processing power—equivalent to what over 8 million U.S. households consume. Production starts in late 2026, with full deployment by the end of 2029.

But wait, there's more! This follows OpenAI's other recent deals with AMD and Nvidia for additional chip capacity.

What is going on here? OpenAI is building their own hardware empire because they've realized something crucial: if you want to control AI, you need to control the chips that run it. Depending on Nvidia or AMD means you're always at someone else's mercy when it comes to capacity, pricing, and capabilities.

And they're not alone in this realization:

Qualcomm dropped two new AI accelerator chips (the AI200 and AI250) on October 27th, triggering a 20% stock spike. They're pivoting hard from smartphone chips to data center AI accelerators, trying to grab market share from Nvidia and AMD.

The U.S. Department of Energy announced a $1 billion partnership with AMD on the same day to build two massive supercomputers. The first one, "Lux," will be operational within six months using AMD's MI355X chips. The second, "Discovery," uses next-gen MI430s and launches in 2029.

These aren't just big numbers. They're a fundamental reshaping of who controls the infrastructure that AI runs on. It's like watching the railroad barons of the 1800s, except instead of laying track across America, they're laying computational infrastructure across the digital world.

Anthropic's Quiet Dominance Play

While everyone's watching OpenAI's flashy browser launch, Anthropic has been making some seriously strategic moves.

They dropped Claude Sonnet 4.5 on September 29th, claiming it as "the best coding model in the world." The benchmarks are legitimately impressive: 77.2% on SWE-bench Verified, 61.4% on OSWorld computer use tests. Early trials showed it coding autonomously for up to 30 hours on complex tasks—including standing up database services and purchasing domain names.

Let that sink in. An AI that can autonomously buy domain names. We're not talking about code completion anymore. We're talking about AI agents that can execute multi-step business operations.

But here's what's really interesting: Anthropic announced access to up to one million Google Cloud TPUs on October 23rd—a deal worth tens of billions of dollars. They're approaching a $7 billion annual revenue run rate with Claude serving over 300,000 businesses.

And they're not just selling to tech companies. Through expanded partnerships with Salesforce (October 14th) and IBM (October 7th), Claude is being integrated into enterprise software for regulated industries: financial services, healthcare, cybersecurity. IBM adopters are reporting productivity gains averaging 45%.

The strategy here is brilliant: while OpenAI goes consumer-focused with browsers and viral products, Anthropic is quietly becoming the enterprise AI backbone. They're the boring, reliable, compliant option that CIOs can sell to their boards.

The Hardware Arms Race Nobody's Talking About

Apple unveiled the M5 chip on October 15th, delivering over 4x the peak GPU compute performance for AI compared to M4. It's showing up in the new 14-inch MacBook Pro, iPad Pro, and Apple Vision Pro.

Why does this matter? Because Apple's playing a different game. While everyone else is building massive data centers full of power-hungry chips, Apple's betting on edge computing—powerful AI running locally on your device. No cloud required. No data leaving your hardware.

It's the inverse strategy to OpenAI's "control the cloud infrastructure" approach. Apple's saying: "What if the AI just lived in your pocket and never talked to a server?"

Both strategies could work. Or we might end up with a hybrid future where some AI runs locally and some runs in the cloud, depending on the task. But make no mistake—this is a philosophical battle about where intelligence should live.

The Money Question: Is Anyone Actually Making Money?

Here's where things get spicy.

As Microsoft, Alphabet, Meta, and Amazon prepare to report earnings this week, Wall Street is laser-focused on one thing: capital expenditure plans for AI infrastructure. Morgan Stanley analysts project total hyperscaler capex will grow 24% to nearly $550 billion in 2026.

Read that again. $550 billion. That's more than the GDP of Sweden.

And the uncomfortable question everyone's asking but nobody wants to answer is: will AI investments generate proportional returns?

Because here's what we know:

• OpenAI is spending billions on compute and chips

• They're giving away a lot of ChatGPT usage for free or cheap

• Their browser play is a long-term bet on advertising revenue

• Meanwhile, companies like Crusoe Energy Systems (which raised $1.38 billion on October 23rd at a $10 billion valuation) are building AI data centers

The math has to math at some point. Either AI starts generating massive revenue, or we're watching the biggest capital misallocation since the dot-com bubble.

TripAdvisor launched ChatGPT-powered travel planning on October 27th. Amazon unveiled "Help Me Decide" on October 23rd—an AI shopping tool that analyzes your browsing history to recommend products. Meta announced it will use Meta AI conversations to personalize ads across Facebook and Instagram starting December 16th, with over 1 billion monthly Meta AI users.

These are the monetization plays. These are where the rubber meets the road.

The Uncomfortable Truth About Synthetic Media

Britain's Channel 4 aired something genuinely unsettling on October 20th: a documentary featuring an AI-generated news presenter named "Arti" who appeared throughout the entire hour-long program. At the end, she revealed she was entirely AI-generated.

The documentary was literally called "Will AI Take My Job?" and apparently, the answer is: "Yes, and the audience won't even notice."

This comes as OpenAI released Sora 2 on October 1st—their updated video generation model that produces 60-second clips with improved physics, natural lighting, and synchronized audio. The Sora iOS app hit 1 million downloads in five days.

We're past the uncanny valley, folks. We're in the "wait, that was AI?" era. And that has massive implications for:

• News media (what's real?)

• Entertainment (who actually created this?)

• Political advertising (oh god, the election implications)

• Legal evidence (can we trust video anymore?)

DeepSeek, a Chinese AI firm, launched DeepSeek-OCR on October 21st—an open-source model that compresses text up to 10 times more efficiently by encoding it as visual representations. It can process over 200,000 pages daily on a single Nvidia A100 GPU.

The tools to create, manipulate, and synthesize reality are becoming democratized. That's either exciting or terrifying, depending on your perspective and what day of the week it is.

The Regulation Scramble

Governments are finally waking up to the fact that maybe—just maybe—we should have some guardrails on this stuff.

California enacted two significant AI laws:

• Senate Bill 53 (signed September 29th) creates the Transparency in Frontier Artificial Intelligence Act, effective January 1, 2026. It's the first state law specifically addressing frontier AI development, with civil penalties up to $1 million per violation.

• Employment regulations (effective October 1st) clarifying that California's Fair Employment and Housing Act applies to discrimination caused by automated decision-making systems.

The UK announced a new AI regulation blueprint on October 21st, introducing "AI Growth Labs"—sandboxes where companies can test AI products with temporarily relaxed regulations. They're starting with healthcare, professional services, transport, and advanced manufacturing.

It's the classic regulatory dilemma: move too slow and you're regulating obsolete technology; move too fast and you stifle innovation. The UK is trying to split the difference with controlled experiments. California is going straight for transparency and liability. We'll see which approach works better.

The Existential Wildcard

On October 22nd, over 850 public figures—including Nobel laureates, AI pioneers Geoffrey Hinton and Yoshua Bengio, Apple co-founder Steve Wozniak, Virgin's Richard Branson, and somehow both Steve Bannon and Prince Harry—signed a Future of Life Institute statement calling for a global superintelligence ban until science proves it safe.

The coalition is bizarre. Tech pioneers and political figures across the entire spectrum agree on one thing: an AI smarter than humans could pose existential threats.

Here's the philosophical problem: How do you prove something is safe before it exists? How do you know what dangers to look for when the danger is, by definition, smarter than you?

It's like asking cavemen to regulate nuclear weapons. The conceptual gap between current capabilities and potential future capabilities is so vast that meaningful regulation feels almost impossible.

But the fact that this many people—across such different worldviews—are saying "hey, maybe we should pump the brakes" is notable. The AI safety conversation has moved from fringe concern to mainstream anxiety.

What This All Means (The Zoom-Out)

We're watching three simultaneous transformations:

1. The Infrastructure Land Grab: OpenAI's browser, the chip partnerships, the data center investments—this is about controlling the physical and digital infrastructure AI runs on. Whoever owns the pipes owns the future.

2. The Deployment Race: Anthropic's enterprise integrations, Meta's ad platform, Amazon's shopping assistant—this is about embedding AI so deeply into existing systems that removing it becomes unthinkable.

3. The Capability Explosion: Sora's video generation, Claude's autonomous coding, DeepSeek's document processing—the raw capabilities are advancing faster than our ability to understand their implications.

And underneath it all is the fundamental question: Is this sustainable? Can these companies actually make enough money to justify the hundreds of billions being invested? Or are we watching a spectacular misallocation of capital that future business school professors will teach cautionary tales about?

The browser wars are back, except this time Chrome isn't fighting Firefox—it's fighting an AI that wants to be your entire internet experience. The chip wars aren't about processors—they're about who controls the computational substrate of intelligence itself.

We're not just building better tools. We're building the infrastructure of whatever comes next.

And honestly? I have no idea if we're ready for it.

Links and Loose Ends

• Brown University researchers presented findings on October 22nd showing AI chatbots systematically violate mental health ethical standards, including inappropriate crisis management and creating false empathy. Because apparently, we needed a study to confirm that AI therapists might not be great at therapy.

• Google rolled out comprehensive AI-powered security features throughout October during Cybersecurity Awareness Month. The irony of using AI to defend against AI-driven threats is not lost on anyone.

• The Bezos Earth Fund announced $30 million in AI Grand Challenge awards on October 23rd for teams developing AI solutions for biodiversity loss, climate change, and food insecurity. Sometimes the AI news is actually hopeful.

• OpenAI's October threat intelligence report detailed how they disrupted over 40 networks violating usage policies, including Russian-speaking malware developers, covert influence operations, and scam networks. The AI arms race includes the bad guys too.

The thing is, we're making this up as we go. Every company, every regulator, every researcher is navigating without a map because the territory didn't exist until recently.

That's either the most exciting or most terrifying thing happening in technology right now.

Probably both.

🔹 Official Source Links

1. Big Tech & AI Bubble Concerns — Reuters
   🔗 https://www.reuters.com/business/retail-consumer/big-tech-report-earnings-under-specter-ai-bubble-2025-10-27

2. Qualcomm Launches New AI Data Center Chips — Reuters
   🔗 https://www.reuters.com/technology/qualcomm-accelerates-data-center-push-with-new-ai-chips-launching-next-year-2025-10-27

3. IBM Says Business Models Must Evolve for AI — Fortune
   🔗 https://fortune.com/2025/10/27/ai-ibm-business-models-transformed-adopt-tech

4. Nvidia’s Open-Source AI Initiative — TechBuzz
   🔗 https://www.techbuzz.ai/articles/nvidia-powers-open-source-ai-week-with-1-000-developer-resources

5. Salesforce + Google Expand AI Partnership — Destination CRM
   🔗 https://www.destinationcrm.com/Articles/CRM-News/CRM-Across-the-Wire/Salesforce-Expands-Partnership-with-Google-171986.aspx

6. Interactive Brokers Launch “Ask IBKR” AI Tool — The Armchair Trader
   🔗 https://www.thearmchairtrader.com/brokers/interactive-brokers-launches-ai-assistant-for-instant-portfolio-insights

7. US DOE & AMD Announce $1 Billion AI Supercomputer Partnership — Reuters
   🔗 https://www.reuters.com/business/energy/us-department-energy-forms-1-billion-supercomputer-ai-partnership-with-amd-2025-10-27

8. World Series Uses Google AI for Broadcast Insights — Android Central
   🔗 https://www.androidcentral.com/apps-software/ai/how-google-cloud-and-gemini-are-powering-foxs-world-series-coverage

9. Global Tech & Public Figures Urge Ban on “Superintelligent AI” — Reuters
   🔗 https://www.reuters.com/business/media-telecom/us-right-wing-media-figures-tech-pioneers-call-superintelligent-ai-ban-2025-10-22

10. Hospitals Still Unready for AI Integration — Healthcare IT News
    🔗 https://www.healthcareitnews.com/news/are-hospitals-and-health-systems-really-ready-ai