In partnership with
Good morning, chip heads.
While you were grilling this weekend, OpenAI quietly lit a fire under the most valuable company on Earth.
Today, we're breaking down OpenAI's first homemade chip and why the company that actually cashes in is neither the one that built it nor the one it's aimed at.
Investors see ANOTHER return from Masterworks (!!!!)
That’s 6 sales in 7 months. 29 all time. And the performance?
16.5%, 17.6%, and 17.8%, net annualized returns on sold works held longer than one year (See all 29 at Masterworks.com)
It’s not from stocks, private equity, or real estate… it’s from contemporary and post war art. Crazy, right?
With Masterworks, you don’t need to be a BILLIONAIRE to invest in multi-million dollar art anymore.
Historically, the segment overall has had attractive appreciation and low correlation to stocks.*
Masterworks targets works featuring legends like Banksy, Basquiat, and Picasso, identifying what they believe to have significant long-term appreciation potential, not just at the artist level but at the level of individual artworks.
As one of the largest players in the art market, with $1.3 billion invested over 500 artworks, they pass critical advantages through to their 70,000+ members to add art to their portfolios strategically.
Looking to diversify your investments in 2026?
*According to Masterworks data. Investing involves risk. Past performance is not indicative of future returns. See important Reg A disclosures at masterworks.com/cd.
Last week, OpenAI pulled the cover off its first-ever chip.
It's called Jalapeño (OpenAI built it with Broadcom), a quiet $1.5 trillion company most people have never heard of.
Broadcom $AVGO ( ▼ 2.41% ) is the company big tech is hiring to design its custom chips.
It already helps build Google's in-house AI silicon, part of the broader chip race we broke down here.
Now, here's the part that should make Nvidia sweat.
Jalapeño is built for inference → the work a chip does every single time you hit "enter" in ChatGPT.
Training is the expensive process of creating a new ChatGPT model. OpenAI feeds the model huge amounts of data so it can learn how to understand and generate answers. This happens when a new model is built, not every time you chat.
Inference is what happens every time you use ChatGPT. You type a prompt, the already-trained model thinks through it, and gives you an answer.
Inference is the electricity bill that never stops coming.
It's the bigger, never-ending half of the AI tab.
And OpenAI says Jalapeño handles that work for about 50% less cost than the chips everyone buys today.
Meaning every answer ChatGPT gives could cost half as much to produce.
They moved fast, too.
A cutting-edge chip normally takes two to three years to go from blank page to factory.
OpenAI and Broadcom did it in just nine months, about the time it takes to make a human.
Potentially making it the most advanced and fastest chip ever designed.
Microsoft $MSFT ( ▲ 1.62% ) has already agreed to buy 40% of the first production run, with chips landing in data centers by the end of 2026.
But one name is missing from that whole story: the chips OpenAI is replacing are Nvidia's.
Nvidia's $NVDA ( ▼ 1.39% ) biggest customers are designing it out of the half of AI that never stops running and Broadcom is handing them the blueprint.
For three years, one idea has propped up Nvidia's $4 trillion price tag: AI runs on Nvidia (all of it) with no way around it.
That's true for training, where Nvidia's lead is still enormous.
But training is the smaller, lumpier slice.
Source: Nvidia
The real money over the next decade is inference: every chatbot reply, every image, every agent task, running every second of every day.
That's the slice Jalapeño goes after.
And the pattern is the tell.
Google $GOOG ( ▼ 0.48% ) already designs its own inference chips. So does Amazon $AMZN ( ▲ 0.4% ). Now OpenAI does too.
The huge companies that buy the most Nvidia hardware on Earth are all building their own exits at once.
Here's the twist most people miss.
They aren't building these chips alone, almost all of them lean on Broadcom to do it.
Broadcom is the arms dealer of custom silicon.
It doesn't have to beat Nvidia in a fair fight, but collects a toll every time another giant decides it wants off Nvidia's meter.
Nvidia sells the gold-rush shovels for digging the mine (training)
Broadcom sells the custom shovels for the part of the dig that never ends (inference)
So the market prices Nvidia as if it owns all of AI compute, when it really owns the front half.
Nvidia has been the AI toll road because almost every company building AI has had to pay Nvidia for the chips powering it.
But now, a second toll road is emerging: another layer of infrastructure that AI companies can’t avoid, and money is already starting to flow there.
Know someone who owns Nvidia and swears it's bulletproof? Forward this to them →