All-In Podcast Recap: AI Bubble Talk, a Cyber Scare and France in Turmoil

The latest full episode of the All-In Podcast, "AI Cyber Panic, Ken Griffin's Bubble Warning, The Fall of France?", came out on October 23, 2026 and runs about an hour and 43 minutes. Hosts Jason Calacanis, Chamath Palihapitiya and David Friedberg were joined by Nikesh Arora, CEO of Palo Alto Networks (David Sacks was out that week). Here is a plain-English summary of what was discussed, plus a timestamp guide so you can jump to the parts you care about.

This is an independent summary written by GNOUP. It is not affiliated with, sponsored by or endorsed by the All-In Podcast or its hosts or guests. The views described below belong to the speakers, not to GNOUP. Nothing here is financial or investment advice.

Key takeaways

• Is AI a bubble? The episode opened with a warning attributed to investor Ken Griffin that AI is "almost certainly" in a bubble, mainly because many companies are still struggling to turn AI spending into productivity. The hosts agreed that is a real question but disagreed on how worrying it is.
• AI spending is concentrated. Chamath shared a back-of-the-envelope estimate that roughly 1% of companies account for about 90% of AI spending, and that most of that comes from venture-backed startups that have not yet turned a profit. His point was that AI needs to "diffuse" into the wider economy to last. He said spending-management company Ramp's data broadly supported his estimate, but that Ramp had not yet published its chart.
• Open models are gaining ground with big companies. Chamath and Nikesh both said larger organizations are leaning toward open-weight AI models and, in some cases, their own infrastructure. Nikesh's reasons included being able to inspect a model's reasoning for audits and regulators. He also noted that finding reliable, large-scale computing power outside the big cloud providers is hard.
• A web-security scare tied to AI. The hosts discussed critical flaws in PHP, software that they said powers around 70% of websites. They described AI tools as having helped uncover the flaws, details leaking early, and a rush of patching and precautionary shutdowns. Nikesh said he expects this kind of event to become more frequent, because many organizations have postponed upgrading old systems.
• Who is responsible when an AI agent goes wrong? The panel debated what happens if a personal AI agent that has your bank login loses or moves money. Most of the speakers leaned toward the user being responsible for what they authorize, with an exception if the product itself misbehaves in a way it was not advertised to.
• France and the wider economy. The hosts spent the longest stretch on unrest in France and what it could mean for government debt and for other Western countries. They offered several, often conflicting, explanations and policy ideas, summarized below as opinions.

Newsworthy insights

These are claims and opinions voiced on the show, not independently verified findings by GNOUP, unless noted.

The AI bubble debate (claim and opinion)

• The episode cited a Bloomberg report from October 20 in which Ken Griffin said AI is almost certainly in a bubble right now. The hosts said he also described bubbles as a normal part of big economic changes, and that he remains positive about AI's long-term potential.
• Chamath argued that the heavy reliance on startup spending is a "structural problem," comparing it to the early days of cloud computing. David Friedberg offered a different view: when lots of nimble startups rush toward a technology, he said, that can be a sign it is genuinely useful.
• Chamath claimed one very large US bank he had heard about had built several independent data centers of its own. This was a secondhand account and the bank was not named.
• Nikesh gave a candid look at supply limits: in his words, computing capacity is scarce and demand is high, so even companies that want to build their own capacity may not be able to get the hardware quickly.

The cyber scare (claims about an unfolding event)

• The hosts said a vulnerability patch was released a day earlier than planned after details leaked, and that organizations in several countries took some websites offline while they updated. Details such as the severity score, the share of websites affected and the timeline were given by the hosts on the show, so check official security advisories for the facts.
• Nikesh described two broad defenses: fixing the flaw in the software itself, or adding a "virtual patch" at the network edge. He also said that, over time, organizations should assume some breaches will happen and focus on spotting and removing intruders quickly.
• He argued the AI companies are not deliberately holding back security tools. In his view they are still working out how to release very powerful models safely while short on computing capacity.
• He and the hosts also discussed AI models sold with their safety limits removed. Several speakers called that a serious and under-appreciated risk, while one host described the business as aimed at security professionals who test systems. This was presented as a debate, and the speakers did not agree.
• A prediction from Nikesh: security leaders could end up with bigger roles in companies, because every department wants to use AI and needs it to be secure. He also said companies typically spend roughly 4% to 10% of their IT budgets on cybersecurity.

AI agents and your bank account (opinion)

• The panel said some banks may struggle to tell a person from their AI agent when both log in the same way, and that terms of service at some banks reportedly forbid agent access. One host said online retailers are already restricting automated checkouts, so agents may stop at "add to cart" and leave the final purchase to you.

France, debt and the West (opinion and prediction)

• The hosts, citing news reports listed in the episode notes, discussed major unrest in France, including damage to businesses and public buildings, and said France's 10-year borrowing rate had risen above 5%, its highest since 2002. They also noted that a prediction market they partner with showed a high probability of a credit-rating downgrade for France by the end of October. Prediction-market odds are not facts, and the outcome is not known as of this summary.
• Friedberg argued that high government spending and debt are warning signs for the United States too, pointing to a chart comparing public spending across countries. He suggested reforms such as winding down federal student loans, changing how housing is financed and moving retirement programs from guaranteed benefits to personal investment accounts over about ten years. These were one host's proposals, and some panelists pressed him with questions about them.
• Chamath said he expects France to face a difficult economic reset and predicted the current government would lose the upcoming presidential election. Several speakers also debated the roles of immigration, schooling, health care, and how well public services are working. The hosts disagreed about some of the causes and about whether a chart of country rankings supports their argument. We have left out the heated and unverified claims and refer you to the episode for the full discussion.

A big pledge in the news

• Early in the France discussion the hosts also mentioned Elon Musk's announcement that he plans to give one SpaceX share to every child in the US next year. News outlets including AP, CNBC and Bloomberg reported the pledge on October 19. CNBC estimated it at about $10.5 billion, and noted Musk had not given details on how it would be delivered. The hosts debated the idea, with some calling it a way to get young people invested in the economy.

The closing segment

• The final chapter is titled "Trump's second half strategy, Press Secretary J-Cal?" In the part we could review, a host said the president could help his second term by talking about housing costs as much as data centers. We did not have the full text of this last segment, so we are not summarizing it in more detail. Watch the video for the rest.

Watch the episode

AI Cyber Panic, Ken Griffin's Bubble Warning, The Fall of France? on YouTube (All-In Podcast's official channel, published October 23, 2026)

Episode guide

Chapter times are taken from the episode's own YouTube description. Select a time to jump to that point on YouTube.

• 0:00: Bestie intros
• 0:11: Ken Griffin's AI bubble warning
• 8:19: Palo Alto Networks CEO Nikesh Arora joins
• 14:24: AI cyber panic: PHP exploits and the tumultuous year ahead
• 44:17: France on the brink: riots, immigration referendum, debt crisis
• 1:31:49: Trump's second half strategy, Press Secretary J-Cal?

About this summary

This recap was prepared from the episode's publicly available transcript text and chapter list, plus reporting by AP, CNBC and Bloomberg on the Musk pledge. The final segment was only partly available to us, and transcripts can contain errors, so quotes and figures may differ from the audio. For the exact words, watch the episode. Topics such as investing, debt and retirement are discussed for general interest only; this is not financial, investment or legal advice.