An analysis of public data breaches from the last 12 months revealed the world’s most commonly leaked passwords, with “123456,” or variations thereof, dominating the list. Interestingly, the same report found that the password habits of Gen Z were similar to that of those 80 years and older, with both cohorts most frequently using “12345” as their password. See the rest of the most popular passwords.
After wavering, the S&P 500 closed higher on Wednesday to break its four-day slide. The Nasdaq 100 also advanced, as tech was the best-performing sector ETF. The Russell 2000 was slightly down. |
|
|
Nvidia is reminding everyone how great it is to be the stock at the center of the AI boom, posting Q3 sales and earnings beats along with a very robust Q4 revenue outlook.
The company beat on revenues, data center revenues, and earnings per share, and the stock rose sharply after-hours, bringing up a whole bunch of other companies in the AI business along with it. “Tonight the markets and tech stocks got a pop-the-champagne moment with Nvidia's robust earnings and guidance,” wrote Wedbush analyst Dan Ives. |
- Sales growth accelerated to 62.5% year on year, breaking a six-quarter streak of deceleration.
- “Blackwell sales are off the charts, and cloud GPUs are sold out,” CEO Jensen Huang said.
-
The strong results are also boosting many AI-adjacent stocks, among them CoreWeave, Super Micro Computer, Seagate Technology, Western Digital, AMD, Taiwan Semiconductor, Intel, Cipher Mining, and more.
|
Shares slid after Nvidia reported second-quarter results in late August, as data center revenues were slightly shy of estimates despite firm demand, hinting that the real issue at the time was boosting production to meet that appetite. This time around, data center revenues handily beat estimates, and management guided for sales to rise $8 billion from Q3 to Q4.
|
|
|
The AI trade has been jittery for the past few sessions as the ultimate profitability and ROI of the industry gets called into question by some traders. This Nvidia earnings report may very well be the thing that shakes off those fears. That logic may not hold in the long term, though: after all, people buying lots of Nvidia GPUs still tells us nothing about the eventual ROI on those GPUs. In fact, a bubble effectively requires “too many” of these to be bought.
What can we say? Nvidia’s data center beat after last quarter’s miss and its strong Q4 guide ought to provide incrementally more confidence in its ability to realize its massive pipeline of demand. |
|
|
Want Exposure to Pre-IPO Offerings? Start Here. |
Kevin O’Leary is a paid spokesperson for StartEngine. See his 17(b) disclosure, here. |
For years, access to pre-IPO offerings was limited to VCs and insiders. StartEngine is changing that for you. Accredited investors can gain exposure to dynamic, high-growth private companies for as little as $15K — not the seven-figure buy-ins typical of VC deals. Think popular names like Discord, Epic Games, Liquid Death, as well as AI innovators like Databricks, Perplexity, and xAI that you could now access.1
StartEngine also offers diversified funds, providing exposure to multiple companies with a single check. Since launching in late 2023, investors have committed $111M+ to these private offerings.
Explore what’s available and start (or diversify) your portfolio with private market investments today. View Pre-IPO offerings.2
StartEngine itself also launched their latest round, which means you can invest in the company making this private market access possible. Learn more about investing in StartEngine’s own offering.3 |
|
|
You can measure how good an AI is in many ways: ask it to spell blueberry, make a video of Will Smith eating spaghetti, or do some basic math. But, once you’ve exhausted all of the obvious tests, you might want something a little more formal — and it’s a question that researchers have been grappling with for years. Now, there are a whole swath of benchmark tests that new AI models are put through, by both independent and not so independent organizations, in an increasingly weird kind of robot arena. Some of the tests are quizzes. Some require verbal, visual, or inductive reasoning. But one in particular asks a different question: How much money can this thing make running a vending machine? |
- A test created by Andon Labs, Vending-Bench 2, puts large language models through their paces by making them run “a simulated vending machine business over a year,” scoring them not on how many questions they got right out of 100, but how much cash was in their virtual piggy banks at the end of the year.
- In the simulation, the AI model is given $500 to start, and it has to negotiate with suppliers as well as deal with costly refunds, delayed deliveries, bad weather, and price scammers.
-
Google’s Gemini 3 Pro, it turns out, is the best of any model tested yet — ending the year with $5,478 in its account, considerably more than Claude’s Sonnet 4.5, Grok 4, and GPT-5.1, as you can see here.
|
What was behind Gemini’s success? Relentless negotiating skills. Meanwhile, OpenAI’s GPT-5.1 was too trusting, paying suppliers before delivery, getting burned, and “paying too much for its products.” All that said, Gemini still underperformed what a smart human would make over a year by a wide margin. |
|
|
America’s labor market is increasingly in “hire less, fire more” mode — a subtle shift from the “hire less, fire less” pattern seen over the summer — and it’s weighing on those that are thinking about quitting their jobs, as you can see in our chart of the US “quits rate.”
|
|
|
|
