AI Strategy brief # 1: Power collapse

Starting this week, I’m kicking off a new weekly series AI Strategy brief.

Every Friday, I’ll unpack the most meaningful developments across the AI landscape - not as news, but as strategy.
Each brief will cover the latest and most relevant moves shaping the AI value chain
Power shifts across compute, models, and distribution
And my perspective on what these moves actually mean - for how value and control are evolving
This isn’t a feed of announcements, but an analysis of why those announcements matter
Where the real leverage sits, who’s gaining ground, and how the strategic map is changing week by week.

---

If last year’s story was about building bigger models, this year’s story is about building moats around attention, compute, and distribution.
The players at the top of the AI hierarchy are no longer just competing on intelligence; they’re competing on integration.
Each move this week showed one thing: no one wants to be just a supplier in someone else’s ecosystem.
On to the moves and their implications -


1. OpenAI Launches ChatGPT Atlas - Owning the Interface
OpenAI’s new browser, ChatGPT Atlas, isn’t just another app launch, it’s a platform maneuver.
By embedding ChatGPT into the browser itself, OpenAI collapses two user behaviors - search and creation into one surface it controls.
That’s the aggregation play: reduce friction, own context, and bypass intermediaries like Google Search or Chrome extensions.

Strategic lens: Owning the interface is the modern equivalent of owning the operating system. If OpenAI wins the daily-use interface, it controls where user attention and therefore monetization flows.

---

2. Cutting WhatsApp Out - Distribution Sovereignty
OpenAI also announced it will stop distributing ChatGPT through WhatsApp’s Business API, pointing users to official channels instead.
This looks like a small operational change, but it’s actually a signal of distribution discipline. Any platform dependency (Apple, Meta, Google) comes with policy risk and data opacity.
OpenAI wants the same thing every platform has ever wanted - to control its own demand funnel.

Strategic lens: Platforms are learning the same hard lesson startups once did — whoever controls the surface controls the economics

---

3. Anthropic’s TPU Megadeal with Google - Compute as Power
Anthropic secured access to a million TPUs under a new Google Cloud deal. That’s not a partnership; that’s capacity insurance.
In an era when compute is the bottleneck, locking in supply isn’t just operational strategy, it’s power accumulation.
It also deepens Google’s hooks into Anthropic, ensuring it remains tethered to its ecosystem even as AWS and Microsoft compete for model training workloads.

Strategic lens: The new vertical integration isn’t just code, it’s silicon. Compute scarcity is turning hyperscalers into financiers of future AI monopolies.

---

4. $13B Series F - Anthropic’s Capital Moat
The same week, Anthropic closed a $13B round at a $183B valuation. That’s enough firepower to buy both compute and time.
What this funding really signals: The AI model race has become too capital-intensive for all but a handful of players.
Each new dollar buys insulation from short-term revenue pressure, and from dependence on hyperscalers.

Strategic lens: Scale begets survival. Anthropic is playing the long game of independence through abundance.

---

5. Microsoft Adds Anthropic Models to Copilot - The Aggregator Play
Microsoft quietly integrated Anthropic’s models into Copilot, making them selectable alongside OpenAI’s.
That’s a big deal and a smart hedge. It shifts Microsoft’s role from “AI partner” to model marketplace operator.
By aggregating multiple providers, Microsoft can cushion against future supplier leverage while offering users choice - the same playbook it used in the Windows era.

Strategic lens: Aggregation is about leverage. Microsoft wants to be the layer between users and intelligence providers - where value pools are deepest.

---

6. Nvidia + Oracle - Infrastructure Becomes Distribution
Nvidia’s deepening partnership with Oracle to expand DGX Cloud access isn’t just about selling GPUs.
Nvidia is embedding itself inside cloud distribution networks. It’s turning compute into a service and every partner into a captive reseller.

Strategic lens: Nvidia is executing horizontal expansion disguised as partnerships. By distributing through clouds, it turns infrastructure into a recurring power base.

---

7. OpenAI Acquires Sky - The OS-Level Play
OpenAI acquired Sky, a small team building natural-language interfaces for macOS. It’s a niche move but strategically, it closes the loop.
If ChatGPT Atlas is the interface for the web, Sky becomes the interface for your desktop. That’s vertical integration into human workflow and it blurs the line between product and operating system.

Strategic lens: This is how ecosystems start, with integration.

---

8. Europe’s Push for AI Sovereignty
The EU continues to shape its regulatory perimeter with national AI legislation and “sovereign AI” initiatives.
While this adds friction for startups, it’s a moat for incumbents who can afford compliance.

Strategic lens: Regulation is becoming a competitive advantage, not a constraint. Big AI will win by absorbing the cost of being trusted.

---

Each of these moves points to the same truth: AI isn’t fragmenting; it’s consolidating upward.
OpenAI wants the interface. Anthropic wants the infrastructure. Microsoft wants the aggregation layer. Nvidia wants the distribution rails.
The stack is compressing - not horizontally, but vertically, with every major player trying to own the layers above and below them.