AI Came Back Richer And Bolder

Today, we’re diving into:

Dear Folks, this week AI just rewrote three of the biggest rules in your industry, from model access to funding to national sovereignty, and none of it is slowing down anytime soon.

Let us unpack exactly what changed, why it matters to your bottom line, and how to turn this disruption into your next competitive advantage before rivals even notice it happened.

Anthropic Just Reopened Its Most Powerful Model.

If you paused your AI roadmap while Fable 5 was offline, you can resume planning. But the rules have changed, and your usage math changes with them.

The Blackout Is Over, But Not Everything Reset

As of July 1, Claude Fable 5 is back online for users worldwide. The US government and Anthropic worked together to clear it for release after a security review.

You will not get unrestricted access right away. A new usage structure comes attached, and it directly affects how you plan heavy workloads this month.

Your Compliance Team Should Know This

Commerce Secretary Howard Lutnick confirmed the government worked closely with Anthropic over two weeks to approve the release and align it with federal oversight standards.

A Security Flaw Triggered A New Layer Of Defense

The relaunch follows a vulnerability reported by Amazon. Anthropic responded by training a new safety classifier built to catch and block the exploit pattern before it reaches the model.

You should expect this classifier to intercept most malicious attempts. Anthropic states it targets the reported bypass method directly and blocks it at the gate.

Watch For False Positives In Your Workflows

Here is the catch your engineering team needs to plan around. Anthropic openly admits the classifier may misfire on legitimate requests, especially during routine coding and debugging sessions.

Blocked requests do not just fail silently. They get rerouted automatically to Claude Opus 4.8, so your workflow continues even if Fable 5 declines the task.

Your Usage Limits Just Got A Countdown Clock

Access through July 7 is capped at 50 percent of your weekly usage allowance for Pro, Max, Team, and select Enterprise plans. After that window closes, usage shifts to credit-based pricing.

What This Means For Your Next Planning Cycle

Frontier model access is becoming a regulatory checkpoint, not just a product launch. Every rollout now carries usage windows, classifier behavior, and government sign-off baked into the timeline.

Build that volatility into your own AI infrastructure planning before your next migration or deployment cycle depends on a model that could pause again.

Europe Just Built An AI Model To Cut You Out Of The Picture

A national government just launched its own AI model, so it would not have to depend on providers like you. That decision says more about the future of your migration strategy than it does about Portugal.

Portugal Just Went Fully Independent On AI

Portugal switched on Amalia this week, its first open-source AI model. Built for European Portuguese, it gives public institutions a foundation they fully own.

It is not a chatbot for citizens. It is a base infrastructure that agencies, universities, and companies can build their own tools on top of.

Why Your Enterprise Clients Should Care

Early deployments already involve the Portuguese Navy, public museums, lesson-planning tools, and citizen service assistants. That is government infrastructure running on domestically built AI.

This Is Not An Isolated Move

France and Germany are already backing homegrown alternatives like Mistral AI and Aleph Alpha. Italy is building a 400 billion parameter open model spanning EU languages.

Europe's largest cloud provider has said it wants to build frontier models from scratch too, sitting alongside Mistral rather than depending on outside vendors.

The Trigger You Cannot Ignore

This summer, the US government briefly cut off access to Anthropic's most capable models before restoring it. European buyers in finance, government, and healthcare felt that disruption directly.

A recent survey found most German companies now believe their reliance on US cloud providers runs too deep. That sentiment is shaping AI policy across the entire bloc.

What This Means For Your Migration Roadmap

Amalia runs on Portugal's own supercomputing infrastructure, including Deucalion and MareNostrum 5. Funding is already locked in through 2027, so this is not a pilot project.

Build Portability Into Every Client Conversation

Clients locked into a single foreign AI vendor now carry visible geopolitical risk. Multi-cloud and multi-model portability is no longer a nice-to-have on your roadmap.

The organizations that plan for model flexibility today will not scramble when the next access disruption happens somewhere else in the world.

An $8.3 Billion Bet Just Exposed What You Are Overpaying For

One AI infrastructure company just raised 800 million dollars to prove open source can beat closed platforms on cost. If your computer bill still stings, this round is a signal you cannot ignore.

The Funding Round That Rewrote The Valuation Map

Together AI closed an 800 million dollar Series C this week, led by Aramco Ventures. The round pushes its valuation to 8.3 billion dollars.

That is a jump from 3.3 billion dollars just over a year ago. Your infrastructure vendors do not grow that fast without enterprise demand pulling them forward.

Why Your Computer Bill Should Worry You

Together AI built its platform around one core complaint from enterprise buyers. Closed AI systems come with astronomical compute costs that quietly erode your margins.

One client reportedly cut inference costs sixfold after moving workloads onto the platform. That is not a rounding error. That is a budget line you are currently overpaying on.

Open Source Just Became A Boardroom Strategy

Enterprise usage of open source models has tripled over the past twelve months. What used to be a developer side project is now a procurement priority.

The Number That Should Get Your Attention

Together AI's annual bookings recently crossed 1.15 billion dollars. That figure comes almost entirely from companies actively walking away from closed AI ecosystems.

Vendor lock-in is becoming a liability you can quantify. Every dollar spent on a closed platform is a dollar competitors are redirecting toward flexible, customizable infrastructure.

The Infrastructure Race Is Just Getting Started

Together AI is not stopping at funding. The company secured commitments for over 500 megawatts of dedicated compute capacity, capitalized independently by strategic investors.

That scale of hardware investment signals long-term confidence, not a short-term trend. Your infrastructure planning should account for a market shifting away from closed defaults.

What This Means For Your Next Budget Cycle

Closed platforms have controlled AI pricing power for years. This round shows decentralized, open ecosystems are closing that gap fast, and closing it with real enterprise dollars.

Before you renew your next AI infrastructure contract, run the comparison. The cost gap this funding round just validated might already apply to your own stack.

Your opinion matters!

Hope you loved reading our piece of newsletter as much as we had fun writing it. 

Share your experience and feedback with us below ‘cause we take your critique very critically. 

Thank you for reading

-Shen & Towards AGI team