BUSINESS

BTS swept all three of their nominations at the 2026 MTV VMAs, securing Song of the Year for “SWIM,” Best K-Pop, and Best Group.

Taylor Swift broke a tie with Beyoncé, becoming the most-honored artist ever at the Video Music Awards, winning first Artist Director Award

Benjamin Satterley, known as PAC, has shockingly passed away at 40 just 24 hours after competing at AEW All Out 2026.

TheStreet aims to feature only the best products and services. If you buy something via one of our links, we may earn a commission.

Why we love this deal

Temperatures across the country are cooling down with fall’s arrival, and while it might not be time to pull out your heaviest winter coats, an extra layer for warmth will soon be required for heading out the door. My favorite layer to wear in early fall is a lightweight sweatshirt. A sweatshirt isn’t as bulky as a hoodie, making it easier to move around and get things done. It’s also superior to a fleece jacket or zip-up in that it doesn’t have any hard metal zippers that could poke you, leaving only the soft and cuddly fabric remaining. 

As you’re transitioning your clothing from summer to fall, consider having this essential layer in your wardrobe, because Amazon’s no. 1 bestseller in sweatshirts, the Rumia Long-Sleeve Crew Neck Sweatshirt, is now on sale for as low as $9. Featuring a beige plaid print, this $9 sweatshirt looks more elevated than you’d expect from the casual style, with a chic pattern that’s perfect for autumn. Despite the more upscale appearance, the sweatshirt is still just as soft and cozy as you’d expect, making it a stylish, comfortable, and affordable addition to your closet.

Rumia Long-Sleeve Crew Neck Sweatshirt, $9 (was $14) at Amazon

Courtesy of Amazon

Shop at Amazon

Why do shoppers love it?

An excellent choice for everyday wear, this sweatshirt is soft and cozy with a bit of stretch, so you’re not restricted as you run to the coffee shop or visit the pumpkin patch. The polyester spandex blend requires no special care, so you can throw it in the wash on laundry day, but it still has the desirable softness you’d expect from a sweatshirt. One shopper raved, “They are so very soft inside and out!! They feel like a hug from a fluffy cloud.”

Related: Amazon has machine-washable slippers with arch support for $17

While this sweatshirt isn’t as thick as a traditional sweatshirt, it’s still thicker than your average long-sleeve shirt. The medium-weight material smooths out bumps while still offering shape for an improved fit. It’s also a few inches longer than most tops, so it would pair well with leggings and form-fitting pants, as well as jeans and joggers. “This is so soft and the perfect fit,” one shopper said, praising the sweatshirt. They wrote, “It’s not oversized, so it’s very flattering, and the length is just right for leggings or jeans.”

Details to know 

Sizes: The shirt is available in sizes small to 3XL.

Material: 90% polyester, 10% spandex. 

Closure type: Pull-on style.

Is it machine washable?: Yes.

A lightweight sweatshirt like this one can be advantageous compared to the plusher styles. It’s very easy to overheat in a standard sweatshirt, and pulling it off can mess up your hair if you’ve gone to the trouble of fixing it. A thin sweatshirt can be worn much longer before you heat up, and those who run cold can wear it comfortably all day.

Shop more deals

Wiholl Cowgirl Cheetah Sweatshirt Tunic, $13 (was $24) at Amazon

Hotouch Plaid Flannel Shacket, $18 (was $20) at Amazon

Ofeefan Blue Plaid Mock Neck Sweatshirt, $10 (was $20) at Amazon

Stay warm and stylish this autumn with the Rumia Long-Sleeve Crew Neck Sweatshirt. Available for as low as $9, the essential fall layer is a fashion-forward option that provides coziness and warmth.

For an artist whose VMA history spans more than four decades, Madonna’s opener functioned as both a celebration and a reminder.

Polo is attracting a new generation. From arena access to Rivals, social sport and heritage, what its resurgence reveals about the awakening consumer.

Looking for help with today’s New York Times Wordle? Here are some expert hints, clues and commentary to help you solve today’s Wordle and sharpen your guessing game.

Mattel is expanding its “KPop Demon Hunters” action figure line with a third wave that includes a pair of demon Saja Boys members and more.

Ron Baron has already earned billions from Elon Musk.

Now the billionaire investor feels the largest portion of that gamble may still be in front of him.

The founder of Baron Capital made a very positive case for SpaceX on a CNBC “Squawk Box” appearance, suggesting that reducing launch costs, on top of Starlink growth, and accelerating artificial-intelligence demand may drastically raise the company’s long-term worth.

Baron said Baron Capital has invested roughly $25 billion in SpaceX and more than $5 billion in Tesla (TSLA). Of the approximately $71 billion in profits he said the firm has generated for clients over its history, roughly $30 billion came from investments associated with Musk.

His newest prediction about SpaceX is far broader.

Ultimately, Starlink could be producing something like $1 trillion per year in sales and $700 billion to $800 billion in EBITDA, Baron said. Based on his assumptions, he said the satellite business itself could eventually be worth something like $14 trillion to $15 trillion.

Those statistics are estimates from Baron, not financial forecasts from SpaceX.

But advancements at both SpaceX and Tesla are providing investors with much more real data points to judge the Musk enterprises.

Ron Baron sees a massive opportunity beyond SpaceX rockets

Baron’s thesis for SpaceX starts out with one critical variable: the cost of reaching orbit.

Reusable rockets have dramatically cut launch costs compared to the economics of traditional, expendable launch vehicles, he said. He expects Starship will drive those costs down even further, potentially making space accessible to businesses for which that wasn’t economical before.

Starship is a fully reusable transportation system designed by SpaceX to carry more than 100 metric tons to orbit in a reusable configuration.

That’s significant because Baron isn’t just predicting more rockets to launch.

It is based on what cheaper launches allow.

The most established example is Starlink. Baron said he had modeled the satellite network growing dramatically over the next decade, with consumer broadband accounting for only part of the potential revenue opportunity.

Government customers, businesses, and mobile connectivity could contribute substantially more, he argued. And increasingly, Baron believes, AI agents could create another wave of communications demand as autonomous software performs more tasks around the clock.

SpaceX’s own current plans illustrate how aggressively it is trying to expand network capacity.

Starship will allow much more network capacity to be deployed per launch, and the company said its next-generation V3 Starlink satellites are expected to provide far more capacity than today’s generation.

SpaceX is also no longer restricted to private-market holdings.

The company priced its massive public offering in June at $135 per share, selling about 555.6 million Class A shares and raising roughly $75 billion. The stock began trading under the ticker SPCX.

That gives public-market investors a direct way to judge whether SpaceX can grow into anything resembling Baron’s enormous long-term expectations.

“We’re going to make hundreds of billions in the next 10 years from Elon,” Baron said.

Related: Elon Musk sends strong message to SpaceX and Tesla investors

The $15 trillion figure isn’t the most ambitious part of his thesis, either. Baron believes orbital computing could eventually create an even larger opportunity.

SpaceX’s AI data-center plan strengthens Baron’s thesis

One theme Baron kept returning to throughout the interview was space-based data centers .

AI companies are demanding larger and larger amounts of computing infrastructure, which requires electricity, cooling, land, and significant capital.

Some of those constraints could eventually be eased by moving computing infrastructure into orbit, says Baron.

He said he expects SpaceX to begin deploying data center capability into space as early as 2027, though he said the timeline could slip. As for the new business, he called it “Star Mind” and said it could one day be bigger than Starlink.

SpaceX has since publicly documented the underlying project, unlike some of Baron’s more aggressive financial assumptions.

SpaceX says StarMind is a system of satellites in orbit that carry AI-compute hardware powered by solar energy. Its initial AI1 satellite design calls for up to 250 kilowatts of peak computing power, and SpaceX says a planned manufacturing facility in Bastrop is meant to support production of thousands of AI satellites starting as soon as late 2027.

SpaceX also says Starship is key to that strategy because the company will need to launch large amounts of computing hardware at a reasonable cost.

Its current planning assumes that the gradual deployment of millions of tons of satellites could result in a huge amount of AI-compute capacity in orbit. These are company ambitions, not guaranteed outcomes, and building such infrastructure will require advances in launch cadence, satellite manufacturing, computing hardware and thermal management.

Heat management is one of the challenges Baron discussed in the CNBC interview

Hardware in space can’t use air-based cooling like ground-based data centers do. Instead, SpaceX’s StarMind design uses radiative cooling, solar power and laser links to tie the orbital compute network to Starlink.

There is an official SpaceX orbital-AI program, which gives Baron’s larger argument a more concrete grounding.

His valuation for Starlink is still very speculative at $14 trillion to $15 trillion.

But the notion that SpaceX wants to be more than a rocket-and-internet company is no longer theoretical.

Musk’s next growth engine could come from an unlikely placeBloomberg / Getty Images

Tesla’s latest autonomy push backs part of Baron’s argument

SpaceX may be Baron’s biggest bet on Musk, but Tesla is still a big part of his portfolio.

Baron said his firm has more than $5 billion in Tesla stock, and he personally owns about $1.5 billion worth of the EV maker.

And he also cleared his stance on the shares.

Baron said he believes “the time to buy the stock is now,” arguing that Tesla’s Full Self-Driving technology is gaining traction.

Some recent data from Tesla also backs up the specific autonomy trends Baron pointed to, but it doesn’t prove his investment conclusion will be right.

Tesla delivered 480,126 vehicles in the second quarter of 2026, including 467,762 Model 3 and Model Y vehicles. The company produced 451,758 vehicles and delivered 13.5 gigawatt-hours of energy-storage products in the quarter.

Tesla’s financials included revenue of $28.24 billion for the second quarter, with net income attributable to common shareholders of $1.11 billion.

The company said its strategy increasingly centers on bringing AI into the physical world through FSD, Robotaxi and humanoid robots such as Optimus.

More Elon Musk:

Elon Musk makes bizarre claims about money, future of AI

Elon Musk sends blunt verdict on the future of humanity and AI

Elon Musk’s startling claim to SpaceX investors

Tesla’s Robotaxi operation has progressed further since that quarterly report.

The current Robotaxi website for the company lists autonomous ride service in Austin, Dallas, Houston, Miami, Orlando and Tampa. Tesla claims the Cybercab is designed for full autonomy with no steering wheel or pedals.

Tesla says Cybercab is now only available to take rides in limited areas of Austin, while Model Y vehicles support Robotaxi service in other markets.

On Sept. 22, Tesla published a specific Cybercab rider guide explaining how customers can hail and ride driverless Cybercab rides via its Robotaxi app.

Tesla is also making strides on FSD (full self-driving) outside of North America.

On its FSD safety page, the European version currently lists the Netherlands, Lithuania, Estonia, Denmark, Belgium, Slovenia and the Czech Republic as markets where FSD (Supervised) has been approved. The system still needs driver supervision and should not be confused with a fully autonomous product.

But Tesla’s push toward autonomy also faces regulatory scrutiny.

The National Highway Traffic Safety Administration opened an investigation Sept. 4 into Tesla’s self-certification of the Cybercab after it began commercial deployment in Austin. The agency said it is looking into whether the vehicle complies with applicable Federal Motor Vehicle Safety Standards.

This is a necessary counterweight to Baron’s enthusiasm.

FSD subscriptions and Robotaxi deployment are growing, but regulatory approval, technological reliability, and consumer adoption are key variables in how valuable those businesses ultimately become.

Baron is not focusing as much on Tesla’s Optimus humanoid robot business.

Musk believes robots could one day be bigger than Tesla’s other businesses, Baron said. But at the moment, Baron places more emphasis on autos, self-driving, batteries, and energy.

That makes his investment thesis somewhat narrower than Mr. Musk’s own vision for Tesla.

But it also connects directly to what Tesla is doing today: the rapid growth in adoption of FSD, the expansion of the Robotaxi service, and the move of Cybercab from development into the real world.

Baron’s Musk bet is becoming an AI infrastructure bet

Baron started wagering on Musk with electric cars and rockets.

It increasingly feels like a gamble on the AI infrastructure.

Tesla is trying to make AI into physical products with autonomous cars and robots.

SpaceX is bringing more satellite connectivity online and building up orbital computing infrastructure that could potentially provide AI models with lots of compute.

Baron believes those trends could come together.

AI agents require connectivity. You need to be connected to network capacity. The AI models need the computing power. And SpaceX has a launch platform that could deliver communications and compute infrastructure to orbit at an unprecedented scale.

There are big expectations.

Baron sees a path to a Starlink revenue run rate of about $1 trillion and a valuation of as much as $15 trillion. He thinks orbital AI computing could be even bigger in the future.

Neither is guaranteed.

To build that future, Starship launch rates, satellite production, customer adoption, and AI-compute demand will all have to increase steeply, and significant engineering and regulatory hurdles will have to be overcome.

Tesla’s own unknowns lie in its efforts to transform ever more capable autonomous-driving technology into a mass-scale transportation business.

Yet the latest twists make one aspect of Baron’s thesis easier to grasp.

His Musk bet is no longer primarily about cars or rockets.

More and more, it’s about controlling the infrastructure required to connect, move, and provide computing power for an AI-driven economy.

And the opportunity is just beginning, says Baron.

Related: The Robotaxi payday Tesla promised owners isn’t coming

There is a version of the Oracle story that sounds like triumph. Cloud infrastructure revenue up 121%. A $664 billion backlog. $30 billion in new AI contracts signed in a single quarter. 300,000 GPUs delivered to customers, and management raising the full-year revenue forecast to at least $90 billion. I covered all of that in my previous piece.

Then there is another version that sounds like a warning. Credit default swaps at record highs. The 2056 bonds yielding over 8% for the first time. A force majeure notice delaying payments on the New Mexico Project Jupiter data center. S&P downgrading the firm to BBB, one notch above junk. 

If that final downgrade comes, $120 billion of Oracle bonds would be automatically removed from investment-grade indexes, according to a Seeking Alpha analysis.

Both versions are simultaneously true. That tension is the Oracle story in 2026, and it matters for anyone trying to understand how the entire AI infrastructure buildout is being financed.

Also read: How much Oracle stock is Larry Ellison using as loan collateral? 

The cash flow picture that is scaring bond markets

Oracle spent approximately $28.5 billion on capital expenditure in Q1 fiscal 2027 against quarterly revenue of $19.35 billion. Free cash flow for the trailing twelve months through August 2026 is -$28.72 billion, according to Alpha Spread.

That number looks catastrophic until you understand what is driving it. Oracle’s operating cash flow remains strongly positive before the massive growth spending on GPUs, power infrastructure, and new data centers. The negative free cash flow is largely the result of discretionary investment against a contracted backlog, not a deterioration in the underlying business.

Now, the bull case: once Oracle slows its growth capex engine, the company could return to generating substantial free cash flow. Its underlying software and support business, while shrinking as a share of revenue, still carries exceptional margins. And the Oracle Cloud Infrastructure (OCI) buildout represents heavy upfront investment to support a multi-year revenue stream.

More AI Stocks:

Citi reiterates its Buy and $330 target on Oracle

Truist says CoreWeave stock could nearly double to $165

Jim Cramer reveals 6 AI stocks to watch in 2026

The bear case is that Oracle has yet to give investors a clear timeline for when free cash flow will turn positive again. 

Management expects fiscal 2027 net cash capex to approach $70 billion, against roughly $90 billion to $95 billion in reported capex, according to the Q1 2027 Earnings Call Transcript. At that scale, Oracle needs sustained access to external financing to keep the buildout going.

The bond market is charging Oracle for that uncertainty. Five-year credit default swaps for Oracle have reached levels far above the investment-grade average, according to a Seeking Alpha report.

Latest Goldman data via Reuters show that AI-related issuers are paying about 115 basis points on new debt versus 78 basis points for the broader investment-grade market. But Oracle’s situation is more acute than the average.

How Oracle is trying to manage the financing problem

The company is not ignoring the issue. Two structural approaches are cutting down its capital burden.

Customer prepayments. Customers increasingly fund hardware purchases themselves through bring-your-own-hardware arrangements. In Q4 fiscal 2026, approximately $75 billion in AI infrastructure contracts were structured primarily through prepayments or BYOH structures. Q1’s new contracts, management said, did not require additional Oracle capital to fulfill.

The Apollo and Blackstone $35 billion AI purpose vehicle I covered when reporting on Broadcom — Oracle is using similar third-party financing structures to keep specific deployments off its direct balance sheet.

If customers fund a large portion of their own infrastructure, Oracle can grow OCI dramatically without proportional increases in its own spending. I think that is the best-case scenario for the capital structure concern.

Oracle Cloud Infrastructure revenue is up 121%, with a $664 billion backlog and $30 billion in new AI contracts signed in a single quarter.Shutterstock

The business transformation that changes how Oracle should be valued

The deeper issue Morgan Stanley has been wrestling with, as I covered in the Sep. 10 earnings note, is that Oracle is becoming a fundamentally different company.

Software revenue fell 3% in Q1 fiscal 2027. The high-margin, asset-light software business that defined Oracle’s economics for decades is shrinking as a proportion of the total. 

Growth is now being driven by cloud infrastructure, which requires significant physical capital for every incremental dollar of revenue.

Oracle’s long-term margin target for Oracle cloud infrastructure (OCI) is above 30%, with high-20% returns on invested capital for mature data centers. If Oracle hits those targets, this capex cycle could look like one of the most profitable infrastructure investments in technology history.

If they are not achieved — if utilization ramps slowly, if energy and GPU costs remain elevated, if customers negotiate down pricing as more supply comes online — the math looks very different.

The $664 billion backlog provides revenue visibility. It does not provide equal visibility into the economics of converting that backlog into cash. 

You see that gap between contracted revenue and actual free cash flow? That is the question Oracle’s bond market is pricing, and the one equity investors need to think carefully about before treating the 121% OCI growth number as the complete story.

Related: Jim Cramer sends strong signal to Oracle stock investors