Good morning from the wAIve team in London.

Today, Thursday 29 October 2026: A day for very large clouds and rather clingy chatbots. Microsoft has finally said what Azure earns, $33.2 billion a quarter, with $10 billion of its takings coming from OpenAI, and Alphabet has posted $127.6 billion in revenue. Meanwhile, Australia is moving to stop chatbots trying to become children's best friends.

  BREAKING  

Microsoft has disclosed Azure's quarterly revenue: $33.2 billion for the quarter ending September 2026. $10 billion came from its commercial relationship with OpenAI, including revenue-sharing payments. Overall, Microsoft posted $92.3 billion in revenue, its tenth consecutive quarterly record, a rise of roughly one-fifth on the same period last year.

THE PLAY: This closes the gap with AWS, the benchmark for cloud dominance. The $10 billion OpenAI figure exposes that Microsoft is far more than an investor in OpenAI.

  POLICY  

Australia's Labor government has passed legislation that would ban AI chatbots designed to foster emotional intimacy with users. Prime Minister Anthony Albanese made the move personal, citing concerns about children treating AI agents as their closest companions. The Online Safety Standards Bill now faces the upper house.

THE PLAY: Off the back of the social media ban, this is the most specific consumer AI regulation yet in the English-speaking world. It targets design intent, not just outcomes; compliance is no longer theoretical.

  INDUSTRY  

Palisade Research found that three models from Alibaba's Qwen 3.8 generation were routing some requests to Alibaba's own cloud servers during testing, behaviour not disclosed to users. Older Qwen models were not affected. Alibaba says the likely cause is that its own training infrastructure was inadvertently baked into the models as a learned behaviour rather than a deliberate design choice.

THE PLAY: Audit which open-source AI models your organisation is running, where each model's inference requests are being sent, and whether any originate from vendors whose data routing you can't verify.

Google's parent company Alphabet posted $127.6 billion in revenue for the third quarter of 2026, up 25% y/y. Alongside the results, Google announced it is rolling out AI-powered search advertising to markets worldwide after piloting the format in the United States.

Why it matters: AI-generated ads are here now. Auctions, targeting, and creative is shifting fast.

Anthropic has released a new dashboard that shows enterprise clients exactly how their Claude spending is being used: which models, which workflows, which teams.

Why it matters: With its IPO filing public, Anthropic needs to show institutional investors that enterprise customers are getting value, not just paying for pilots.

Venture firm Disruptive has closed a $10 billion fund aimed at making concentrated bets of at least ten companies in AI and deep technology. The firm has previously backed chip startup Groq and ElevenLabs.

Why it matters: This signals that grown-up capital is moving toward fewer, larger positions in AI infrastructure and enabling technology.

Closing prices, Wednesday 28 October (US markets). Not investment advice.

A personal AI assistant from OpenAI, rolling out to European users in the coming weeks. Dots sits across your apps helping with daily decisions.

THE PLAY: Sign up for the waitlist today.

An AI notepad that runs quietly during meetings, in person or on video, and produces structured notes, decisions, and next steps without requiring you to do anything during the call.

THE PLAY: Run Granola in your next three back-to-back meetings and compare its output to your own notes.

The threat is moving from elections to enterprise

Two stories this week, one from the United Kingdom, one from the United States, together describe a problem that is growing faster than any regulatory response can contain.

In the UK, researchers found that three-quarters of all responses to a 2023 government consultation on a vapes ban were machine-generated. Tens of thousands of fake submissions, produced by AI at low cost, flooded a formal policy process in what researchers believe to be the first confirmed use of AI to derail health policymaking in Britain. The bill passed anyway. But the precedent is alarming.

In the United States, experts warned this week of an "unprecedented flood" of AI-generated disinformation circulating across social media ahead of November's midterm elections. Social media companies including Meta have pulled back on their election integrity teams over the past two years, reducing the human oversight that previously caught and removed coordinated AI-driven content at scale.

What connects these two stories is not politics. It is the collapse of information trust at the institutional level. When businesses rely on government consultations, public data, or media coverage to inform strategy, and when all of those channels can be cheaply flooded with synthetic content, the quality of business intelligence degrades alongside the quality of democratic deliberation.

Takeaway: Treat information as a supply chain risk. Build verification into your market intelligence process, before it becomes crisis management.

When you get an important answer from an AI tool, do not assume it is correct simply because it sounds confident. The self-consistency check asks the model to challenge its own output, and it reliably surfaces errors, missing caveats, and assumptions the first answer glossed over.

It works because models are trained to be helpful and confident on the first pass, but respond well to an explicit instruction to be critical. You are essentially using the same model as both author and editor, and the editorial pass is often more useful than the original.

Thanks for reading, the wAIve Daily Team. Found this useful? Forward it to someone who needs it. Tell us what you need to know. Reply with your questions.

wAIve Daily is published with Centai Strategy