Oil, Chips, and Builders: Monday's rebound, Google's new TPUs, and the skill that pays

Monday opened with a bounce. After a brutal week, Wall Street came back swinging, oil fell hard, and Google dropped two new AI chips on the same morning. Bitcoin, meanwhile, sat the party out. Let me walk you through what happened, what it means, and what you can build with it today, starting from zero.

The market bounced, but it is still arguing with itself

Last week ended ugly. Thursday brought a 4.36 percent rout, the worst session since April 2025, and Friday only steadied the week instead of repairing it. The S and P 500 closed Friday at 7,376.86, up 0.31 percent. The Nasdaq closed at 24,579.42, up 1.09 percent. The Dow eased 0.46 percent to 51,285.73.

Monday opened higher across the board. The Dow rose 540.52 points, or 1.05 percent, to 51,826.25 at the open. The S and P 500 rose 59.90 points, or 0.81 percent, to 7,436.76. The Nasdaq rose 200.72 points, or 0.82 percent, to 24,780.15.

What moved it? Oil fell hard. Brent dropped about 5 percent to 93.38 dollars a barrel, and United States crude fell nearly 3 percent to around 88 dollars. Reports said President Donald Trump signaled further United States attacks on Iran were unlikely if peace negotiations kept making progress. Cheaper oil eased inflation fears, the 10 year Treasury yield eased to 4.833 percent, and the VIX dropped more than 9 percent.

Chip stocks led the way. AMD jumped about 11 percent after an analyst upgrade, and TSMC rose over 6 percent. But Nvidia opened down over 8 percent. Its board said Sunday it has opened an independent investigation after learning that chief executive Jensen Huang was sought out by a foreign intelligence effort aimed at extracting the company’s technology secrets.

Here is the question that matters. Is this a real turn, or a bounce inside a bigger argument? Friday gave us a strong jobs report, with employers adding 165,000 jobs in October against about 65,000 expected, and the jobless rate easing to 4.1 percent. That strength pushed traders to mark the odds of a December rate hike up to about 78 percent. Tuesday brings the October consumer price report at 8:30 AM Eastern, with economists expecting headline inflation of 2.7 percent for the year. That number is the week’s verdict.

Bitcoin sat out the party

While stocks bounced, crypto did not get the memo. Bitcoin trades at 74,820 dollars, down 1.3 percent in the last 24 hours. Ethereum trades at 2,125.66 dollars, down 4.5 percent. The total crypto market cap sits at 2.52 trillion dollars, down 3.16 percent in 24 hours. Bitcoin dominance is 59.6 percent, and the Fear and Greed index reads 37, which is fear.

Read that plainly. Stocks rose on cheaper oil and a jobs beat. Crypto fell anyway. That split tells you crypto is trading on its own worries right now, not riding the stock market’s mood. Fear at 37 means the crowd is nervous, not panicked. If you are starting from zero, the lesson is simple. Do not chase a bounce you do not understand. Learn the asset before you buy the asset.

Google just showed you where the AI money is going

The biggest AI news of the last 48 hours came from Google on Monday morning. Amin Vahdat, Google’s senior vice president and chief technologist for AI and infrastructure, announced two new systems in Google’s eighth generation of Tensor Processing Units, the custom AI chips Google designs in house. The new systems are TPU 8tx and TPU 8i Max. With the TPU 8t and TPU 8i introduced in April, Google has now introduced 4 TPU systems in a single year, which Vahdat said is a first.

Here is what each one does, in plain terms.

TPU 8tx is built for confidential computing. It runs on the same silicon as the TPU 8t training chip, paired with 5th generation AMD EPYC processors as hosts. Its job is to keep your data and the model’s weights protected and isolated even while the AI is actively processing them. Think of it as a locked room where the work happens, and nobody outside the room can peek in. Google says it delivers up to 2 times lower latency than TPU 8t. It is aimed at government agencies and regulated companies, like banks and hospitals, that want to run AI on sensitive data without handing that data to anyone else.

TPU 8i Max is built for memory. It packs 2 TPU 8i chips into a single package, doubling the compute, the on chip memory, and the high bandwidth memory speed of a single 8i, while fitting into the same rack footprint. Its job is long context work: AI agents that chew through huge documents, whole codebases, and long videos without forgetting the beginning by the time they reach the end.

Two honest caveats. Google gave no prices and no ship dates for either system, and Bloomberg reported both are due in 2027. Every performance figure is Google’s own. No independent benchmark exists yet, and no customer was named for either system. Google said its TPUs now serve nearly 100 customers, including frontier labs, trading firms, retailers, and universities.

The chip war is no longer just about raw speed. It is about trust and memory.

Why does this matter to you? Because the company that can run your private data safely, and remember a million words at once, wins the next wave of AI work. Nvidia still runs most AI training today, but Google just told the world it intends to take a bigger piece of that business with its own silicon.

What can you actually do with this today?

Here is the practical part. In October, Google open sourced TorchTPU, a tool that lets developers run existing PyTorch code and Hugging Face models natively on Google’s TPUs without rewriting code. That is the door, and it is open right now.

Do this today. Open Google Colab, which is free, and start a new notebook. Change the runtime type to TPU. Then load a small open model from Hugging Face, something like a text summarizer, and run it on the TPU. You will learn 3 things in one afternoon.

1. What a TPU actually feels like to use.

2. How PyTorch code moves from a regular computer to an accelerator.

3. Whether your idea needs a giant model at all, or whether a small one does the job for free.

That is the skill that pays. Everyone online is arguing about which chip company wins. Almost nobody is learning to run a model on the hardware that already exists. The person who can take a model, run it cheaply, and put it in front of real users is the person who gets paid. The arguing is free. The building is not.

The uncomfortable truth

Markets bounce. Chips get faster. Headlines change every morning. None of that builds anything for you.

The uncomfortable truth is that most people will watch this whole AI wave from the sidelines, reading about other people’s chips and other people’s money, while the people who win are the ones who opened a notebook and ran something today. You do not need a 74,820 dollar Bitcoin. You do not need a data center. You need one afternoon, one free tool, and the willingness to look stupid while you learn.

Start from zero. Stay in the game. Never quit.

Keep building with me

• Join my free Skool community: https://www.skool.com/henry-jenkins-5224

• My website: https://henryljenkins.com

• LightofLife.global

• Twitter @Nevaquit

• YouTube @HenryJenkins3

• Subscribe to Henry’s Substack: https://jenkinsh.substack.com/subscribe