This fortnight the story moved beneath the models themselves, down to the infrastructure and the politics that decide who benefits from AI. China launched a rival governance bloc on the same day Japan switched on a national AI factory built for industry. The markets that have funded the boom are flinching, while Google's own workers have marched against the job cuts that AI is helping to justify. For a manufacturing leader, these are headlines that shape where your future models come from and how much the tools sitting on top of them will cost.
1. China Launches a Rival AI Bloc, and India Sits Between Two Camps
At the World AI Conference in Shanghai, China formally launched the World Artificial Intelligence Cooperation Organisation, a 29-country group headquartered in Shanghai and backed by nations including Indonesia, Brazil, Malaysia, South Africa and Russia. President Xi Jinping urged countries to keep AI from being dominated by any single power, and pushed open-source models as the route to wider access. Analysts read the move as an effort to shape global AI rules at the UN before the US and its allies set them.
What this means for you:
- A second centre of AI influence is forming, built around open-source models and Global South membership.
- Capable models you can run yourself, without a US vendor's terms, are becoming a real option.
Action tip:
Add at least one strong open-source model to the set you evaluate, alongside the US commercial tools you already test. Running a capable model on your own infrastructure lowers your exposure to any single country's export rules or pricing. Treat model origin as a supply decision, and keep a working alternative from a different bloc.
2. A Font Only Humans Can Read Exposes an AI Blind Spot Worth Watching
A San Francisco developer released Ghost Font, which hides text inside moving dots that the human eye reads through motion but that leading AI models fail to decode from any single frame. It is a curiosity rather than a security product, and better vision models will likely crack it. The more serious signal came alongside it, when researchers showed they could hide instructions inside web page styling that AI assistants would follow but a human reader would never see.
What this means for you:
- Humans and AI do not take the same meaning from the same page.
- That gap is now being used deliberately to hide instructions inside ordinary files and pages.
Action tip:
Before you let an AI agent read and act on outside content such as supplier emails or web pages, put a human check on any action that moves money or data. Ask any agent vendor how they defend against hidden or injected instructions. Keep the agent's permissions narrow, so a deceived model cannot reach your critical systems.
3. Google's Own Staff March Against AI-Driven Job Cuts
More than 100 Google employees rallied at the company's California headquarters as the Alphabet Workers Union delivered a petition signed by around 4,500 staff to CEO Sundar Pichai. They asked for guaranteed severance and for voluntary buyouts to be offered before any forced layoffs. They also want an end to performance ratings they say rest on quotas rather than merit. The action lands as Alphabet posts near-record results, with quarterly revenue up 22%, even as tech firms cut staff while raising AI spending.
What this means for you:
- The workforce backlash to AI-linked layoffs has reached the industry's most profitable companies.
- Cutting roles while visibly investing in AI carries a reputational and morale cost.
Action tip:
When AI takes over part of a role, move the person to higher-value work before you consider removing the role itself. Say plainly how you will use AI and what it means for jobs, because silence breeds the fear that drives good staff out. A structured operating layer such as AIFM helps you redesign work around AI rather than simply cut headcount.
4. Japan Builds the First National AI Factory Aimed at Industry
Japan's government, industrial leaders and NVIDIA announced what they call the world's first national AI infrastructure for physical AI, built by Noetra with 13,750 NVIDIA Vera CPUs and 27,500 Rubin GPUs across 140 megawatts of capacity. It anchors Japan's FRONTia project, which pools the country's manufacturing data to build foundation models for robots and factory automation. The pretrained models will be shared broadly with Japanese firms, and the wider strategy targets 30% of the global AI robotics market by 2040, a stake put at $133 billion.
What this means for you:
- Governments are now building sovereign AI capacity aimed squarely at manufacturing.
- This gives firms in those countries cheaper access to models trained on real industrial data.
Action tip:
Find out whether India or your key markets run a shared industrial AI or robotics programme you can join or buy into. Models trained on sector data tend to beat general tools on factory tasks, and public programmes often subsidise access. Position now to use industrial foundation models as they arrive, rather than building everything from scratch.
5. The AI Trade Just Had Its First Real Scare
Chip stocks fell hard worldwide despite strong earnings, as investors began to doubt whether AI-linked spending can keep growing at its recent pace. TSMC reported second-quarter revenue up 33.7% to $40.2 billion, yet its shares still slipped, and the main US semiconductor index dropped about 12% over the week. UBS projects that the big cloud providers' capital spending will slow from 76% growth this year to 6% by 2028, which is the fear now pricing into the market. Asian and European tech shares led the sell-off, with Japan's Nikkei falling into correction.
What this means for you:
- The spending boom that has driven AI tool prices down is being questioned for the first time in months.
- If the cloud providers pull back, falling prices and rapid upgrades could slow or reverse.
- This is a caution about pace rather than a collapse.
Action tip:
Do not budget on the assumption that AI tools keep getting cheaper every quarter. Lock in pricing where a workflow already pays back, rather than waiting for a better deal that may not arrive. Keep your AI spending tied to proven returns, so a market wobble does not strand an over-committed project.
That’s all for this fortnight. The pattern this fortnight sits under the surface of AI, in the infrastructure and the money. China and Japan are each building the base layer on their own terms, while investors have started to ask how long the spending can run. The human signals matter too, from workers resisting cuts to a font that reminds us machines and people still see the world differently. The firms that gain will read these shifts as supply and workforce decisions, not as distant news.
Until next time,
The AI-First Mindset Team
About Us
At AI-First Mindset, we help leaders bridge the gap between knowing AI and using it. Our work includes hands-on workshops and operating frameworks designed for companies that want to embed AI into workflows without relying on large institutional vendors.
To explore how we can help, contact us at aifirstmindset.ai.