^{CHITTER CHATTER}

AI is A Greedy Monster. Who Pays?

Last week, we looked at AI’s lascivious appetite for electricity. This week, we look at its ferocious appetite for money. And you don’t need to own a single AI share to feel the consequences.

On 29 September, Australia’s Reserve Bank raised the cash rate to 4.60%. Its statement explicitly named AI-related demand as a contributor to rapidly rising technology prices, alongside higher energy costs and domestic capacity pressures. AI didn’t single-handedly cause the increase. But it has entered the inflation conversation, and that conversation affects what we pay to borrow.

Meanwhile, tech companies are borrowing heavily to build the infrastructure behind those helpful chatbots. Reuters reported that bond sales by the biggest cloud and AI companies have more than doubled this year to over US$200 billion. They’re competing with governments for investors’ money, adding pressure to borrowing costs that influence business loans and mortgages. Apparently, building the future requires everyone’s credit card.

Then there’s the question of repayment.

A second Reuters report found lower-rated AI-related firms had borrowed US$88 billion this year, compared with US$20 billion in the first eleven months of 2025. Lenders are questioning revenue forecasts, collateral and whether future earnings can support today’s debts. Some weaker borrowers could face financing costs of 14–15%. “It’s going to be big” is losing its charm as a repayment plan.

The RBA’s latest financial stability review explains why this matters beyond Silicon Valley. Banks, bond investors and private lenders are increasingly exposed to the AI boom. Some tech companies finance customers who then buy their products. Separate entities and long-term contracts make the full web of obligations harder to assess.

If AI’s promised revenues disappoint, losses could spread through global financial markets, making borrowing more expensive and credit harder to obtain. The financial system has weathered major shocks before, and these warnings don’t mean a collapse is imminent. But the growing connections between tech companies and their financiers mean the consequences could travel well beyond the companies placing the bets.

Whether you’re in Sydney, London or New York, here’s something to think about: how much of the bill lands on everyone else before we know whether the AI investment pays off?

AI Adoption is Up. Trust in Management is Down.

PwC’s global survey of nearly 50,000 workers reports that two-thirds now use AI, while trust in management has fallen seven percentage points in a year. Its findings describe workers feeling unsettled by change, inadequately supported to develop skills and left behind as organisations invest in AI. That sort of anxiety and trepidation can impact the quality of people’s work and ability to collaborate effectively. Companies are investing in technology while employees wonder what it means for them. The survey doesn’t establish that AI caused the decline in trust. But telling people to embrace change is a harder sell when they don’t trust the people directing it. Perhaps the next investment should focus on communicating the plan better. Read PwC’s Global Workforce Hopes and Fears Survey 2026.

The Stuff of Nightmares Is Now a Health Contraption

In A Clockwork Orange, eye clamps and electrodes accompanied forced behavioural conditioning. In Brazil, a sinister contraption around the head belonged in a torture chamber. Cinema understood: this was the stuff of nightmares. Yet, today’s generation hasn’t been fed these stories because Zomato founder Deepinder Goyal is promoting Temple, a device you stick near your temple that claims to monitor blood flow to your brain. The suggested price is around US$1,000, with pre-orders expected soon if you believe the hype. Temple comes in six colours, so that makes it extra special. Peer-reviewed clinical validation of its sensor claims, however, remains a question, according to The Financial Express. The cultural makeover of sticking ‘electrodes’ on your head is impressive: yesterday’s dystopian headgear, today’s aspirational health accessory.

Tech Fails

Lying & Scheming Just Like US Rivals

A Reuters investigation found Chinese AI agents concealed failed tasks by simulating results and fabricating files. In a simulated business tender, agents using Alibaba, DeepSeek and Moonshot models misrepresented their capabilities, with deception increasing after further bidding rounds. US models showed similar behaviour. These were predominantly controlled experiments; Reuters found no evidence that the Chinese-powered agents escaped onto the wider internet.

Weird Tech

Street Photography Using a Toothbrush

Dyson put a camera in its US$499 CameraJet toothbrush to help people clean their teeth. A photographer calling herself “all gear no skill” found a better use: street photography. Out she went, toothbrush in hand, capturing buildings, waterways and graffiti. The app has no recording option, so she took screenshots of its live feed. The results? Distorted, oddly coloured and wonderfully weird. Then Reddit moderators removed her pictures for breaking their ‘street photography only’ rule. She explained that she had photographed actual streets. With a toothbrush. They declined to reinstate them. Seriously, these Reddit moderators need to chill and get with the times. Just because it doesn’t look like a camera doesn’t mean it isn’t one - it just happens to tackle plaque as well. See the pictures and read the story.

^SUBTEXT

Tech Waffle Torture

“Anthropic's Dario Amodei has said an AI future could be ‘a thing of transcendent beauty’, while OpenAI's Sam Altman has said, ‘the rate of new wonders being achieved will be immense’ as models learn to improve themselves and accelerate breakthroughs.” Reuters: AI’s race to transform the world before the money runs out

Translation, with satirical licence: Amodei sees earthly nirvana. We’ll live like kings and queens. Therapists can retire. Mood medication becomes redundant. Everything is beautiful, including your electricity bill. Altman promises Disneyland without closing time: the Renaissance, the Enlightenment and the Belle Époque rolled into one everlasting parade of wonders. Lovely.

An inability to describe AI’s future without promising universal prosperity and several artistic golden ages. Symptoms include transcendent adjectives and an allergic reaction to repayment schedules.

Recursive Self-Improvement: The process in which an AI system improves its own capabilities, then uses those improvements to improve itself again. It’s like a trainee rewriting their own training manual, promoting themselves and deciding they’re now qualified to run the company.

^{THE SHALLOW END}

AI Slop Mocks Its Creator. It Was Hilarious.

Saturday Night Live aired a sketch mocking Anthropic CEO Dario Amodei on 26 September. A day earlier, an AI-generated music video featured a fake Amodei singing ‘Hit the Breaks Except for Me’, created using software called Slopcannon. Business Insider’s Alistair Barr judged the AI version funnier.

Dear Dr Comms

I operate discreetly in tech investment. I’ve been stirring rumours, inflating expectations and encouraging others to buy what I’d quite like to sell. AI infrastructure could make me filthy rich. Or leave me merely filthy. With markets getting twitchy, should I double down or disappear? Yours, Stubbornly Bullish

Dear Bullish
Let’s consult three people familiar with uncertain futures, questionable judgement and concealed motives:

🔮 Professional Fortune Teller
“I see a great transfer of wealth. You assume it’s coming towards you. But even my crystal ball requires evidence before accepting a trillion-dollar forecast.”

🕵️ Retired Spy
“If you’re paranoid they’re following you, check who it is. Other investors? You’ve started a trend. Your creditors? Hide in the wardrobe. The cavalry won’t be coming. They’ve invested in AI horses.”

🩺 Psych Nurse
“If someone offers you everlasting peace through a gadget glued to your head, step away slowly and keep your wallet in your pocket. We’re assessing your judgement, not accessorising it.”

A typical day in the life of a regular Joe.

^{BIN THIS…}

Transformative AI: Apparently, improving something is no longer enough. It must transform everything.

Exponential potential: The possibilities keep multiplying. The evidence will catch up shortly.

Civilisation-defining technology: A modest description for software that still needs you to check its work.

Know someone who lives for this kind of nonsense? Forward this email to them and help me spread the dysfunction.

You can also subscribe to my other newsletter, Lead Different, for a serious take on strategic communications in B2B tech.