Why I predicted volatility in the AI space before most

The AI market is horrendously down. If you look at my post history I have been sounding the alarm bell on it for around 3 weeks here's my opinion.

The S&P 500 closed October 30 down 1.82% at 7,580.07 and the Nasdaq Composite down 2.86% at 25,437.00, their worst days since June 5. The PHLX Semiconductor Index fell 7.20%, Intel 28.58% on its own supply and PC problems, AMD 10.88%, Nvidia 4.22%. Coverage tied it to doubts about AI spending. An equal-weight S&P 500 fund only fell 0.79% and six of 11 sector funds went up. The S&P 500 closed 3.05% below its record, still up 10.73% this year.

[...] AI is useful but not everyone's going to benefit 1000x productivity from it as it was originally pitched. Plus it will always be copied by opensource Chinese models. Every model upgrade is getting better in smaller and smaller increments. No AGI coming soon IMO. Some will think its FUD. Also, with the way the Claude and OpenAI CEOs have been behaving, nationalization is a real possibility. I'm really hope it's not but this marketing push to call it "Superintelligence" and hype up every unsubstantial projects points to it going down. Get ready.

Source: LinkedIn post, October 7, 2026, 7:01 AM ET, 23 days before the drop

All the LinkedIn posts and comments here went up October 7 to 16. Six of the eight posts carry an undated Edited label and comments show no edit marks, so that wording is proven as of my October 22 data export. It all matches the live posts on October 30. My LinkedIn shows as Bill B.

Source: First example, live on LinkedIn, checked October 30, 2026 · Latest example, live on LinkedIn, checked October 30, 2026

Good enough for most users

By throwing too many resources at once, and overhyping AI technology, it is in my opinion that we will see a huge bubble burst. AI is the most profound technology of all time but the hype went massively overboard. Firstly the assumption that AI would make everyone more productive is false. A small minority use AI and an even smaller minority pay for AI, so clearly a very small minority is using it to its "full potential." AI is the lamp that attracted the money of institutions believing we would have little sentient beings that would piece together how to produce infinite goods creating unlimited prosperity. Making humans obsolete at the workplace and currency meaningless, as well as retail, who aren't sure what AGI even is. This lack of understanding has created this interesting phenomena where both institutions and retail are essentially looking for outcomes only present within Science Fiction movies. Neither side has fully educated itself and we've been told countless times that AGI would be here by 2027. [...] AI is here to stay, but don't be surprised if we overinvested in a technology still in its infancy stage.

Source: LinkedIn post, October 12, 2026, 7:09 PM ET, 18 days before the drop

Opensource models are obviously less capable but good enough for most users. Kimi is overrated, Opus 5.5 is much better but it does not matter because even models from 2 years ago are suitable for moderate users. AI is a fulltime job and it takes a long time to become proficient. In my experience it is much more prone to burn-out than traditional software was, so a lot of people try to build something but abandon. So the amount of heavy users, which these companies rely on is likely to stay flat, decrease, or rise modestly from here. Which ties into the next point.

Money holes

Frontier models can be money holes that trade many times above revenue. OpenAI and Anthropic are valued many times their revenue which may be why OpenAI decided not to IPO this year and Anthropic might decide to halt as well.

OpenAI last raised at $852 billion in March, Anthropic at $965 billion in May. Reported annualized revenue was about $50 billion at OpenAI in September and $65 billion at Anthropic in July, measured differently. Anthropic filed publicly October 26. Altman said no IPO this year and cited safety and readiness. Blaming valuation is a guess.

Source: CNBC, March 31, 2026 · Orrick, May 29, 2026 · Bloomberg on Yahoo Finance, October 8, 2026 · CNBC, August 17, 2026 · Anthropic S-1, October 26, 2026 · Forbes Australia, September 13, 2026 · Fortune on YouTube, September 12, 2026

Everyone says AI will be bigger than the internet, and I agree with that statement. But... What they fail to mention is what happened between when the internet first started and what it is today... the Dotcom bubble. And never were the investments as large in the internet as they are in AI today.

Source: LinkedIn post, October 13, 2026, 6:05 PM ET, 17 days before the drop

It feels bad betting against the economy it really does, but I've been considering this for the past 3-ish months.

Source: LinkedIn comment, October 13, 2026, 6:10 PM ET, 17 days before the drop

AI CEOs

Sam Altman promised us a team of PhD-level experts in our pocket in 2025 with the release of GPT-5. Elon musk said that "money won't matter." Dario's doomsday predictions. Really horrible marketing.

Yes, AI is very cool and I work with it constantly, but the robot utopia we were promised isn't going to happen anytime soon. Revelations require time, not only pouring gasoline over pension fund money. I'd imagine some sort large correction soon, but I'll still be here building regardless.

Source: LinkedIn comment, October 12, 2026, 7:13 PM ET, 18 days before the drop

Altman's PhD line came in August 2025. Musk's, in July 2026, was about 2036. Amodei warned in May 2025 that half of entry-level white-collar jobs could go. He says he isn't a doomer.

Source: Fox Business, August 7, 2025 · The Singju Post, July 29, 2026 · Fortune, May 28, 2025 · Dario Amodei, January 2026 · The Hans India, June 11, 2026

Funny money

As an AI-Native person, in my experience products that do small parts of larger projects I've built have received billions in funding. We call these AI-Wrappers. This is in part due to a lack of understanding investors have of the technology, and rapidly evolving space.

It's a big bubble, take a look at the projects being funded. If you've worked with AI heavy you'll know what I mean.

Source: LinkedIn comment, October 7, 2026, 7:06 AM ET, 23 days before the drop

Critics call Cursor a wrapper and it raised over $3 billion in 2025. It isn't one of my projects. The label is disputed, since Cursor has its own model.

Source: AI Eating the World, October 3, 2026 · TechCrunch, June 5, 2025 · Cursor, November 13, 2025 · CNBC, June 16, 2026

Chips

GPUs are very expensive due to scarce supply and perceived demand. Letting NVIDIA hold leverage against companies like Amazon, Meta, and leading them to buy in huge multi-year deals, in my opinion. Staff at these same companies were encouraged to use tokens, and an absurd amount. Deepseek was revolutionary because it showed AI could be optimized for fewer GPUs. Further lessening perceived demand.

Meta's multi-year NVIDIA deal and Amazon's commitment to over 3 million GPUs disclose no financial terms, so no minimum purchase is documented. Token leaderboards at both companies have since come down. DeepSeek's V3 report says it trained on 2,048 GPUs. Nvidia still hit $5 trillion nine months after the DeepSeek scare.

Source: NVIDIA, February 17, 2026 · Motley Fool, August 28, 2026 · Fortune, April 9, 2026 · TheStreet on Yahoo Finance, May 31, 2026 · DeepSeek V3 report, December 27, 2024 · CNBC, January 27, 2025 · RNZ/CNN, October 30, 2025

Last week I wrote that AI has been overhyped and that we’re heading for a bubble. Today I bought puts to back that up. All of them expire September 17, 2027, which gives the thesis about 11 months to play out. Exited my $MU position which went from 100B -> 1T in two years. I view this as an insurance policy. [...] The reasoning: CoreWeave and Oracle are the most leveraged bets on the AI buildout, because they borrowed heavily to build data centers. Nvidia sells the shovels in this gold rush. Meta is spending the most on AI while being the least clear about how that spending makes money. I’m not betting against AI. I think it’s here to stay. I’m betting that the money spent on it got ahead of what the technology can actually do today. Not financial advice. This is just me putting my own money behind my view. [...]

Source: LinkedIn post, October 13, 2026, 3:06 PM ET, 17 days before the drop

Meta isn't the biggest spender in dollars. CNBC puts its 2026 capital spending at $135 billion to $145 billion. Amazon's is about $220 billion.

Source: CNBC, October 29, 2026 · NBC News, October 30, 2026

Citadel's Ken Griffin said investors are "almost certainly" in an AI bubble on October 20. That was 13 days after my October 7 bubble comment and a week after my October 13 Dotcom post. He stayed bullish long term. At Davos in January he'd already said the hype helps justify the spending.

Source: Decrypt, October 21, 2026 · 24/7 Wall St., October 21, 2026 · Barchart on Yahoo Finance, October 22, 2026 · Fortune, May 18, 2026 · IBTimes UK, January 29, 2026

Musk weighed in on X on Sunday, October 25, 12 days after my October 13 CoreWeave post. He said over 200 neoclouds (AI cloud firms) would face "guppy-level death and consolidation." He named no neocloud, and said SpaceX had closed $40 billion of debt to buy GPUs.

Source: Fortune, October 26, 2026 · 24/7 Wall St., October 26, 2026 · Reuters via The Star, October 26, 2026

From the close the day each stock was named to October 30: CoreWeave -29.48%, Meta -19.35%, Nvidia -10.45%, AMD -10.21%, Oracle -8.75%, Palantir -8.17%, Broadcom -5.30%. The strikes I posted sit far lower. These are stock moves, not my result on the puts.

Source: Yahoo Finance CRWV data, October 30, 2026 · Yahoo Finance META data, October 30, 2026 · Yahoo Finance NVDA data, October 30, 2026 · Yahoo Finance AMD data, October 30, 2026 · Yahoo Finance ORCL data, October 30, 2026 · Yahoo Finance PLTR data, October 30, 2026 · Yahoo Finance AVGO data, October 30, 2026 · Positions named on LinkedIn, October 15, 2026

Look at my coreweave prediction which I published 3 days ago. People have overestimated the demand for AI and data centers. [...]

Source: LinkedIn post, October 16, 2026, 4:22 PM ET, 14 days before the drop

The reporting doesn't back up the demand part. The worry was payoff and financing, and Nvidia's CEO called neocloud demand very strong.

Source: CNBC live blog, October 22, 2026 · CNBC live blog, October 30, 2026 · CNBC, October 30, 2026

Zuckerberg is a genius, but I'm betting on $META going down in price. Why? They have no clear vertical. They do both software, hardware, and AI. They want to be a "everything" company. The best restaurants have 2 things on the menu, not everything you could possibly imagine eating. [...]

Source: LinkedIn post, October 16, 2026, 8:09 PM ET, 14 days before the drop

People aren't using it to its potential

In my experience conversations on linkedin have revolved around AI emailing and writing for almost 4 years, people aren't using it to its potential.

How can AI be the greatest invention but most overhyped as it currently stands? An unprecedented amount of money went into this. Like a lot. Huge promises were made in short timelines. Today I saw something about AI being able to cure cancer. I'm sure it will be able to one day, but we're getting ahead of ourselves. Most aren't using AI to its fullest capacity and only a few use it outside of personal therapist/email writer. It's harder to go up than it is down. Also China is making a point by using these same frontier models to produce copies opensource. They're saying "you invest, we will copy." Not to mention the scrutiny Anthropic and OpenAI will be under during IPO process.

Source: LinkedIn post, October 13, 2026, 4:09 PM ET, 17 days before the drop

NBER found writing, emails included, was about 40% of ChatGPT work messages in July 2025, the top use. The China quote above is my paraphrase.

Source: NBER working paper, September 2025

After the AI bubble bursts I expect the amount of job postings for AI jobs to go down. A lot of people will call it a "fad." And it will be dismissed as something akin to NFTs when in reality, it's the best invention to date. [...]

Source: LinkedIn post, October 14, 2026, 11:21 PM ET, 16 days before the drop

Disclosure: This is my personal opinion and it isn't investment advice. I don't know your situation, and nothing here tells you to buy or sell anything. As of October 31, 2026, I hold put options on CoreWeave, Nvidia, Oracle, Meta, AMD, Broadcom and Palantir. Those puts make money when the stocks fall. I may sell them at any time without telling you. I bought Micron stock around two years ago and sold it by October 13, 2026. These views are mine and not those of any client or firm I work with.

About the author

Bill Baumgarth

Founder of Capital Catalyst. I build and run revenue systems for sales teams.

Read my CV and selected results