For the past three years, most conversations about AI focused on the models.

Which one was smartest? Which wrote the best code? Which generated the best images? Which company would reach AGI first?

Those questions still matter.

But they are no longer the main event.

A different battle is emerging underneath all of it, and it may determine who leads the next decade of AI.

The fight has shifted from intelligence to infrastructure.

Consider what happened over the past week.

SpaceX agreed to acquire Anysphere, the company behind the AI coding platform Cursor, in an all-stock deal valued at about $60 billion.

OpenAI expanded its Daybreak cybersecurity initiative and launched Patch the Planet, an effort to help open-source maintainers find and fix vulnerabilities in widely used software.

Micron announced a strategic agreement with Anthropic to support the memory and storage infrastructure behind future AI workloads.

The Senate Judiciary Committee unanimously advanced the NO FAKES Act, a bill designed to protect people from unauthorized AI replicas of their voice and likeness.

And Anthropic faced federal restrictions that forced it to limit access to some of its most advanced models.

Viewed individually, these look like separate stories.

Viewed together, they reveal something much larger.

The AI industry is becoming vertically integrated.

The companies positioning themselves for the next decade are not simply trying to build better models. They are competing for control of the infrastructure that makes those models possible.

That includes chips.

Memory.

Data centers.

Developer tools.

Cybersecurity systems.

Energy.

Identity protections.

And increasingly, regulatory access.

Start with Cursor.

Cursor is not just another coding app. It sits directly inside the software development workflow. Developers use it to write, edit, review, and ship code.

That matters because the place where software gets created may become one of the most important layers in the entire AI economy.

If you control the developer workspace, you influence how software is built.

If you influence how software is built, you gain leverage over everything downstream.

That is why the SpaceX deal matters. It is not just about owning a popular tool. It is about owning a position in the production layer of modern technology.

The Micron and Anthropic agreement tells another side of the same story.

For years, AI has been discussed as if it were weightless. Models lived in the cloud. Users interacted with them through a browser window. The hardware underneath remained mostly invisible.

That illusion is fading.

AI is physical.

Every token generated by a model depends on memory, compute, networking, storage, and power. As models become more capable and agents become more persistent, those physical requirements become more important.

The companies that secure access to memory, chips, and energy are not just reducing costs.

They are building strategic advantage.

Cybersecurity is entering a similar phase.

OpenAI’s Daybreak expansion and Patch the Planet initiative point toward a future where AI does not merely identify threats. It helps repair them.

For decades, cybersecurity was primarily a human-driven discipline supported by software tools.

That model is changing.

The software itself is becoming an active participant in defense.

This shift matters because the same agentic systems that introduce new risks may also become essential to protecting the infrastructure they depend on.

Then there is the Anthropic story.

The broader lesson is simple: frontier AI access can no longer be assumed.

For years, users and companies treated cloud-based AI as if it would always be available through a web link. That assumption is starting to look fragile.

If access to advanced models can be shaped by export controls, national security concerns, litigation, or geopolitical pressure, then AI availability becomes a business continuity issue.

That is a major shift.

Companies that rely on one model, one vendor, or one cloud pathway are not just making a technical choice.

They are accepting infrastructure risk.

The NO FAKES Act adds one more layer.

The bill is not about compute or data centers. It is about identity.

But in the AI era, identity is infrastructure too.

Voice, likeness, image, and reputation are becoming assets that require legal protection. As synthetic media becomes cheaper and more convincing, the ability to prove what is real, authorized, and human becomes part of the trust layer of the internet.

That is why this moment feels important.

The infrastructure of AI is no longer limited to servers and semiconductors.

It now includes developer platforms, memory supply, cyber defense systems, identity protections, export rules, and governance frameworks.

The first phase of AI was about capability.

The next phase is about control.

Who controls the chips?

Who controls the developer tools?

Who controls access?

Who controls the security layers?

Who controls identity?

Who controls the rules that determine who can use the most powerful systems?

Those questions may matter more than who builds the most impressive model.

History offers a useful parallel.

The companies that benefited most from the internet were not always the ones that invented the underlying technologies. Many succeeded because they controlled critical pieces of infrastructure.

Cloud platforms.

Mobile operating systems.

Search engines.

Payment networks.

Distribution channels.

AI may follow the same path.

The winners of the next decade may not simply be the companies with the smartest models.

They may be the ones that own the foundations underneath them.

Busy executive. Packed calendar. An inbox that never empties.

Everyone is talking about AI agents. Every week brings another promise of an assistant that can do it all.

Catch is the real deal.

A smart, proactive AI admin agent focused solely on taking administrative work off your plate.

It schedules meetings, triages your inbox, drafts emails in your voice, resolves conflicts, sends follow-ups, and handles the countless small tasks that consume your day.

Available wherever you work — Gmail, Outlook, Slack, WhatsApp, and even over the phone.

No setup. No training. No learning curve.

Catch learns how you work, takes action when it's confident, and keeps things moving without constant supervision.

From swamped to sorted in seconds.

Get started with Catch and have your assistant ready before your next meeting.

Test Catch for free

Developer tools are becoming strategic assets.
The SpaceX and Cursor deal suggests that the software creation layer is becoming a major battleground in AI.

AI access is becoming infrastructure risk.
The Anthropic restrictions show that frontier model availability can be shaped by government policy, export controls, and national security concerns.

Cyber defense is becoming machine-speed.
OpenAI’s Daybreak and Patch the Planet efforts point toward a future where AI systems help find, validate, and repair vulnerabilities across critical software.

Memory is becoming a strategic bottleneck.
Micron’s agreement with Anthropic reinforces that AI progress depends on physical infrastructure, not just model design.

Digital identity is becoming protected infrastructure.
The NO FAKES Act signals that voice, likeness, and personal identity are becoming central legal concerns in the generative AI era.

June 29 through July 2 | San Francisco

One of the most important technical AI gatherings of the year, focused on agents, developer tooling, applied AI engineering, local deployment, and the infrastructure behind real-world systems.

August 1 through 6 | Las Vegas

One of the world’s premier cybersecurity conferences. Given the rise of AI-assisted threats and autonomous cyber defense, this year’s event should offer important signals about where security is heading.

August 4 through 6 | Las Vegas

A major AI conference focused on enterprise adoption, real-world implementation, generative AI, agentic systems, governance, and business outcomes.

September 15 through 17 | Santa Clara

A highly relevant event for this issue’s theme, focused on compute, data centers, AI data movement, physical AI, and the infrastructure layer powering the next phase of AI.

September 29 through October 1 | San Francisco

A major applied AI conference bringing together builders, researchers, founders, and leaders focused on moving AI from research into production.

The biggest technology shifts rarely happen in plain sight.

Subscribe to Gritletter for weekly insights on AI, business, technology, and the forces shaping what comes next.

Gritletter.co