Last week was rough if you own tech. The Nasdaq fell 2.1 percent on Friday alone and about 4.5 percent for the week, the main chip ETF lost around 10 percent, oil is back above $100, and the VIX is sitting above 22. The S&P 500 was down about 2 percent. That's the backdrop for the first big week of AI earnings, and it changes what "good" has to look like.
The calendar
Tonight, Tuesday, Netflix reports after the close, along with GE Aerospace, GM, 3M and RTX earlier in the day. Wednesday after the close is the main event: Meta first, then Tesla about an hour later with its call at 5:30 p.m. Eastern. Procter & Gamble reports Thursday morning. Next week brings Microsoft and Alphabet on the 28th and Amazon and Intel on the 29th, though those dates are from calendar services and I'd confirm them on each company's investor page before planning around them. The Fed is in its quiet period and decides rates on the 28th too, so next week is going to be loud.
What I'm listening for
Meta's revenue is expected around $63 billion, up about 23 percent from a year ago. The number itself will probably be fine. The question analysts have been asking since the spring is whether the ad business is getting more efficient because of all the AI spending, or just getting bigger because the ad market is bigger. If Meta raises its capital spending forecast again without a clearer line to the ad numbers, I expect the stock to get hit even on a beat, because that's what happened in a market that just lost 4.5 percent in a week.
Tesla already told us the deliveries: 486,532 for the quarter, well above the roughly 461,000 estimate. So the delivery beat is priced in. What isn't known is the margin on those cars, and whether the robotaxi and robot talk gets any closer to revenue. Estimates for earnings per share are scattered between 30 and 70 cents, which tells you how little agreement there is.
Here's my honest position. I own broad index funds. I already own Meta and Tesla and every chip name through them, in proportion to their size, and I'm not buying or selling either around a Wednesday afternoon press release. Two weeks ago I wrote that a single number about OpenAI's revenue moved the whole chip sector. This week has three or four of those numbers. If you're trading around them, you're betting you can read an earnings call better than the people whose job it is. I can't, and I've stopped pretending.
The quieter story: humans between the agent and the money
While the market argued about capex, the people who regulate banks spent the week drawing a line around AI agents, the software that doesn't just answer questions but goes and does things, like moving money. Four things happened in about five days.
On October 15 the Financial Stability Board published its final guidance on AI in finance. The part that matters is what it calls Sound Practice 10: human oversight should scale with how risky and how autonomous a use is, and for the higher-risk cases a firm needs risk-based triggers that require a human to approve before an agent executes a transaction. The June draft used the phrase "synthetic employees" for these agents. The industry hated it, and the final version dropped the term, but not the idea.
Also last week, the Financial Times reported that Citigroup and Bank of America joined JPMorgan in limiting third-party AI agents' access to consumer accounts. Bank of America said plainly that it wants a human to approve money leaving an account, and that it hasn't worked out who's liable when an agent authorizes a payment. JPMorgan talked about "rogue agents" in the thousands. And on Monday the UK's Financial Conduct Authority said it's planning a framework for what it calls agentic finance, covering how an agent proves who it is, how it gets delegated authority, who's accountable, and how it executes.
One thing I need to correct from a summary I was sent. It said the Monetary Authority of Singapore announced "Safeguards for Agentic Finance" as the theme of its 2026 Global FinTech Hackcelerator. It didn't. Monday's release names 18 finalists working on three problems set by GXS Bank, Julius Baer and Zurich Insurance, with a demo day on November 18 and S$80,000 per winner. The safeguards work is real, but it's a white paper MAS put out in July, not this week's news. Details like that are how a true story turns into a wrong one.
On the other side of the ledger, the money keeps coming for AI that reads your bank data. Prism Data, a New York company that scores creditworthiness from transaction history instead of a credit score, announced a $20 million Series A led by Story Ventures on Monday. It says it works with three of the five largest US fintech lenders and six of the ten largest digital banks. So the direction is clear: more AI deciding whether you get a loan, and more rules saying a person has to sign off before AI moves your money. Both at once.
What I'd actually do this week
If you hold index funds, nothing. If you hold individual tech stocks and Wednesday's calls make you want to act, wait until Friday. The first reaction is often wrong, and the week after earnings is when the real repricing happens. And if a bank or an app ever offers to let an AI agent pay your bills on its own, read what the FSB just wrote. The regulators want a human in that loop. So do I.
I'm not a financial advisor, and this isn't advice for your situation. It's how I'm thinking about mine.
Verified October 19, 2026. Earnings calendar, consensus figures, Tesla deliveries and last week's market moves from Gotrade's weekly outlook and MarketBeat. Tesla's call time from Tesla Oracle. FSB guidance from Forkast and Payment Expert. Bank restrictions from The Paypers, citing the Financial Times. FCA plans from PYMNTS. MAS finalists from the MAS release. Prism Data from FF News.
Comments
Be kind and stay on topic.
No comments yet. Be the first.