🪙 97 Cents. That's The Story.

The next Apple may already exist. Insider sources say Elon has spent two years building a secret device inside Tesla's facilities — one he claims will be "10x bigger than the largest product in history."

There's just one problem: the company is private, and unless you know Elon personally, you can't buy a single share. That was true until Guardian's research team found three public ticker symbols sitting in the launch supply chain.

Click here to see all 3 tickers, free of charge.

You won't hear these names on CNBC — Wall Street hasn't published a word on the connection. But when the launch hits September 21, that quiet ends.

Some are already calling this the biggest opportunity since AI. For anyone who missed Apple before the iPhone, this may be a second look at that kind of setup.

Get the 3 ticker symbols here.

Unlock the free briefing now.

A million AI tokens now costs 97 cents.

That's around 750,000 words of machine-generated text. The whole King James Bible, twice, for less than the price of a coffee.

Goldman Sachs thinks that's a problem for your portfolio.

Analytics firm Silicon Data tracks what the market actually pays for large language models. Its benchmark just hit a fresh low of $0.97 per million tokens.

In August alone, that index dropped 29%.

No bottom yet. It just keeps going.

Both open-weight and proprietary models got cheaper over the last two weeks, but Silicon Data says the closed models did most of the cutting. The expensive stuff is discounting hardest, which is usually what happens when somebody is scared.

Goldman's Delta One desk laid out the mechanics in a note that made the rounds on social media via Zerohedge.

Frontier labs are now punching each other in the wallet. Meta just dropped Muse Spark 1.3, which is benchmarking uncomfortably well. OpenAI is rumoured to be days away from launching something called Astra.

Every new competitor compresses the cost per token a little further.

Meanwhile, inference is drifting toward cheaper and more local hardware. Goldman's take on charging by the token: "I struggle with per token pricing as a durable end state model."

That's a bank politely asking whether the entire business model has a shelf life.

Cheap AI everywhere sounds bullish. More users, more usage, more everything.

Goldman's warning is that volume has to grow faster than price falls, and right now it might not be.

If the cost of producing an AI output collapses quicker than demand for outputs expands, you end up with too many GPUs chasing too little revenue. The desk's phrasing was blunt: "periods of compute oversupply become plausible."

Unless the next wave of model releases delivers a genuine leap in capability, the industry has built a lot of very expensive capacity for a product that keeps getting cheaper.

The AI trade has quietly become the equity market. A big slice of the S&P's valuation now rests on the assumption that AI revenue scales up as fast as the infrastructure bill does.

Token prices falling 29% in a month is the first real stress test of that assumption.

Worth watching: whether the hyperscalers start talking about consumption volume instead of pricing power on the next round of earnings calls. Companies change what they measure when the old number stops looking good.

Ready to stop using AI as a search engine and start using it as an income engine?

The Hustle's "200+ AI-Powered Income Ideas" is your free playbook for turning the most overhyped technology of our time into actual cash.

Inside you'll find:

Subscribe free today and unlock the full guide. The people already cashing in aren't smarter than you, they just started earlier.

Get The Free Guide

Ford wants to sell 100,000 electric trucks in year one.

Exactly one company has ever done that in America. It's the company Ford is trying to beat. (Any guess before the big reveal later?)

The truck is called the Fathom, it starts at nearly $30,000, and orders open early next year.

That price puts an electric pickup in the same aisle as a boring midsize sedan. Not a toy for early adopters. A default option.

Ford's pitch is that the buyer isn't an EV shopper at all. They're going after people currently driving gas and hybrid vehicles, which is a much bigger pond than the under-$40,000 EV crowd.

The spec sheet is aimed squarely at that person:

So the target is ambitious, but not delusional. Just historically unprecedented.

For scale on what "unprecedented" means: Tesla sold roughly 357,000 Model Ys in 2025 and over 190,000 Model 3s. Nobody else has cracked 100,000 units on a single EV model in a US year. Ever.

Which raises the obvious question. What is Tesla doing while Ford aims at its record?

Tesla is showing off the Cybercab Thursday at an invite-only event in downtown Austin, 4:45pm local time (5:45pm ET). Wall Street has been bidding the stock up going in.

It's a two-seater with no steering wheel and no pedals, powered by a single AC three-phase permanent-magnet motor mounted up front. Charging is expected to be inductive, meaning wireless, because a robotaxi fleet can't exactly plug itself in.

No lidar. No radar. Cameras, onboard compute, and millions of miles of driving data.

Musk has called multiple sensors "a fool's errand", arguing that when lidar and cameras disagree, you've added confusion rather than safety. He's said that's why Waymos avoid highways, and why Tesla switched off its radars.

Bold claim. Here's the number that makes it matter.

Stripping out sensors and using a simplified "unboxed" build process gets a Cybercab down to $25,000 to $30,000.

Compare that to the competition:

That's a robotaxi costing roughly a fifth of a Waymo. Morgan Stanley's Andrew Percoco told clients that minimising sensor complexity, if it actually works reliably, could hand Tesla a serious cost and scale advantage and reshape the economics of autonomous transport.

On paper, Tesla wins the maths by a mile. On paper, everyone wins.

Two versions of the same bet are being placed at once, and both hinge on $30,000 being the magic number.

Ford's bet is boring and testable. Build a cheap electric truck, sell it to people who've never wanted an EV, and find out within twelve months whether the price was the barrier all along.

Tesla's bet is cheaper per unit and far harder to prove. Cost advantage means nothing until the cars drive reliably with nobody watching. Thursday is a product reveal, not evidence.

Ford's target gets graded on a delivery number. Tesla's gets graded on a safety record. Very different homework.

The U.S. Treasury is buying back up to $12.5 billion of its own debt today.

Yes, the government is purchasing its own IOUs. Before they're even due.

Here's why that's less weird than it sounds.

Treasury buybacks are basically janitorial work for the bond market.

Older bonds go stale. Nobody wants to trade them. They sit there gathering dust while everyone piles into the shiny new issues.

The Treasury steps in, mops up the unloved stuff, and keeps the world's most important market from seizing up.

Today's operation is the routine version. The bigger news is what's coming.

Treasury is at least doubling its liquidity-support buybacks for long-dated bonds, from a $2B cap to at least $4B per operation.

That covers the 10-to-20 year and 20-to-30 year sectors. Starts September 9, runs through November 4.

So what does that tell us? The long end of the curve needs more help than it used to.

You don't double the size of your liquidity backstop because everything's going great.

Yields eased ahead of Friday's nonfarm payrolls report.

Gold climbed about 1% to $4,430 an ounce as the dollar and yields both slipped.

Markets are pricing a 63% chance of a 25-basis-point Fed move in September.

Treasury liquidity is the thing nobody thinks about until it breaks. When it breaks, everything breaks with it. Mortgage rates, corporate borrowing costs, your portfolio.

Doubling the backstop at the long end is the Treasury quietly acknowledging that heavy borrowing plus elevated yields is straining the pipes.

Nothing is on fire. Somebody is checking the smoke alarms.

And Friday's jobs number lands in a market already leaning hard one way on the Fed. If payrolls surprise, the repricing will be loud.