Five items that matter to a small business, with sources. News moves fast and vendor announcements are vendor claims — check the linked source before acting on any of it.
United Airlines revised its Contract of Carriage on Wednesday, November 11, barring unauthorised third-party AI agents from making, modifying or servicing a ticket, or buying any service or product — and reserving the right to cancel, without notice, any reservation an AI agent made, modified or serviced. The ban covers United’s website and app as well as travel agencies and aggregators, and United says it is the first US carrier to draw this line. AI assistants can still show United fares and schedules, but the customer has to complete the purchase. United cited data security — handing account credentials to a third-party assistant can expose sensitive information and trigger unintended purchases or use of MileagePlus miles — and federal check-in rules, since online check-in requires a passenger to personally acknowledge rules on prohibited items. Chief executive Scott Kirby has said he does not want to create “another middleman” between the airline and its customers. Bookings an AI agent made before November 11 will stand.
Why it matters: This is the first hard line a major seller has drawn against agentic commerce: the airline wants the customer relationship to run through it, not through someone else’s assistant. For a small business, the message cuts both ways — an AI assistant that books travel on your behalf may soon find closed doors, and the rules about who may buy what in your name are being written now, industry by industry.
USA Today: United Airlines bans personal AI agents from booking flights →
Federal Reserve Vice Chair for Supervision Michelle Bowman said on Tuesday, November 10, that a “large correction in equity valuations” is among the main risks to the US economy if returns on AI investments fall short of the high expectations baked into prices. Speaking at the Council on Foreign Relations in Washington, she put numbers on the buildout: AI-related debt — corporate bonds, private credit and project finance — is on track to exceed $800 billion this year, and other obligations, including long-term leases, contract commitments and residual-value guarantees, carry potential payments estimated at more than three times that amount. She also warned that off-balance-sheet liabilities and equity stakes by technology suppliers in the companies buying their products make counterparty risk harder to assess, and that AI revenues, while rising quickly, do not yet cover the cost of maintaining existing models and building new data centres. On inflation, she said October’s consumer price data suggest 12-month core PCE inflation “likely moved up further to 3.3 percent”.
Why it matters: The Fed is now naming the AI trade as a financial-stability risk — not calling a crash, but warning, in conditional terms, that if the revenue does not catch up with the cost, the correction hits share prices first and the funding for the buildout second. For a small business, the practical takeaway is to plan on AI prices and availability being less predictable than the vendors’ roadmaps suggest.
Bloomberg Law: Fed’s Bowman Warns of Stock Correction If AI Bets Don’t Pan Out →
CoreWeave, the “neocloud” that rents out computing power for AI workloads, reported third-quarter results on Tuesday, November 10: revenue of $3.66 billion, up 168% on a year earlier, but a net loss of $709 million and $846 million in quarterly interest expense. The company said it expects about $9 billion in customer prepayments and deposits over the next three months — more than $3 billion of it tied to new contracts for capacity it has already built — and that it will not use the unsecured debt market to fund the business at current yields. Instead it will borrow against specific customer contracts, and it is shifting more capacity to shorter-term deals that command higher prices: more than $40 million per megawatt a year, against $20–$25 million for new four- and five-year contracts. Its revenue backlog stands at $132.5 billion.
Why it matters: The company renting out AI compute now asks its customers to fund the buildout up front — a sign the expansion costs more than debt markets will cheaply finance. For a small business, the nearer signal is pricing: short-term capacity is getting more expensive, and that cost eventually lands in the price of the models you rent by the token.
TechCrunch: CoreWeave turns to customer prepayments, shorter contracts to fund AI growth →
US businesses spent more on AI in October than in any previous month, but only slightly more: spending rose 1.65% from September, according to the November edition of the Ramp AI Index, published November 11 and built from payments by more than 85,000 American businesses on Ramp’s corporate card and bill-pay platform. Paid adoption reached 56.8% of businesses; Anthropic leads on adoption at 44.8%, up 1.1 points, while OpenAI held flat at 40.6%. The top 1% of companies by AI spending now pay a median $10,199.20 per employee per month, up 9%, while the median firm pays $16.32 — and Ramp notes the top 1% of businesses drive 80% of the revenue OpenAI and Anthropic collect in its data. Engineering teams still drive 72% of classified AI spending, but customer support’s share jumped from 0.8% in June to 13.3% in October, the fastest-growing use case in the index.
Why it matters: Adoption is widening but the money is concentrating: a few big spenders pay nearly everything, while the typical business spends about $16 a month per employee. And the growth story has moved — it is no longer engineering driving the spend, but customer support, where AI agents are taking on real front-line work.
Ramp: November 2026 Ramp AI Index: New Fronts in the AI Trade →
Most organisations cannot say what AI actually costs them, according to a new snapshot report from the Capgemini Research Institute, published Thursday, November 12: 59% of executives say their organisation’s AI-related spending over the past 12 months was significantly higher than expected or budgeted — 24% above expectations on average — and around one in five organisations (21%) overshot their AI budget by more than 30%. Only 56% are fully confident they can forecast how AI costs will change as adoption scales, and only one in five calculates “AI Yield”, the verified value of an AI-assisted outcome after its full safe-delivery cost. More than three in five organisations expect to adopt a multi-model strategy over the next 12 months. The report is based on a September 2026 survey of 500 CXO-level executives at organisations with annual revenues above $1 billion, across 10 countries and 13 industries.
Why it matters: AI costs are proving harder to predict than the vendors’ pitches suggest — and most businesses learn that after the money is spent. For a small business, the lesson is to count the full cost — oversight, rework, governance — before committing to a tool, not after the first invoice.
This day has a permanent page: November 12, 2026. Every day is kept — see the News archive. The News page always shows the latest day.
← Latest day · News archive · Previous day: November 11, 2026
For what the AI companies and their leaders are saying on social media — claims and opinions, not verified news — see The Social Wire →