This Week in AI: NVIDIA Profits Defy Bubble Fears
For Nov. 21, 2025: Google grabs attention with Gemini 3, U.S. ups nuclear power investments after tech, ChatGPT for teachers.News

AI's impact on our lives at home and work is already undeniable, with billions of users spread across OpenAI's ChatGPT, Meta AI, Google Gemini, Anthropic's Claude, and many more. The technology has upended tech, of course, but it's also made significant impacts on medicine, education and entertainment. Its reach stretches from world-class weather prediction models to new interactive toys for kids.
But on the financial side, the business world is locked in a perpetual debate about whether the AI industry's profits will ultimately reflect its impact, or whether it will amount to a new expected feature for the products we already buy. The will-it-won't-it debate gets particularly heated every few months, when public companies including Microsoft, Amazon, Google parent Alphabet, and chipmaker NVIDIA report their quarterly earnings to investors.
Just as with past earnings reports, the pessimists (often called "bears") appear to have lost out to optimists (often called "bulls"), particularly as NVIDIA announced record sales of $57 billion for the period between July and October, up 22% from the previous quarter, and up 62% from a year ago.
"Compute demand keeps accelerating and compounding across training and inference — each growing exponentially," said Jensen Huang, NVIDIA's CEO, in a statement with the report. "We’ve entered the virtuous cycle of AI."
Of course, NVIDIA isn't the only company seeing green in the AI race. While startups like OpenAI have yet to turn a profit, mainstays like Amazon, Alphabet and Microsoft have credited the technology with helping boost their businesses. They've also said that in aggregate they plan estimated spending of more than a quarter-trillion dollars to meet computing demands from their customers. The law of large numbers though suggests that at some point, any successful product will eventually hit a plateau
"There’s been a lot of talk about an AI bubble," Huang told investors this week, according to the WSJ. "From our vantage point, we see something very different."
Read more: NVIDIA DGX Spark: Everything You Need to Know
Gemini 3 grabs attention
Google may have had a head start in the AI wars, but it's struggled to keep pace with companies like OpenAI, Anthropic, and others whose AIs have wowed with features for generating code, music and more. This week, Google sought to retake its lead with Gemini 3, which the company says is its biggest release since Gemini 2.5 back toward the beginning of the year.
Google says Gemini 3 is especially good at coding, and emphasized that point through a new desktop app called Antigravity. The new tool, which is free to use while in testing, is designed to work with the code on a computer, rather than relying on a git repository like Google does through its Jules web-based app. Google said that in addition to working with Gemini 3, Antigravity will also support Anthropic's Claude Sonnet 4.5 and OpenAI's GPT-OSS.
So far, reviews have been largely positive, with influencers calling it among the best out there. "The last time we saw a capability jump of this magnitude was the release of GPT 4 in March of 2023," YouTuber Theo Browne said in a recent video praising the launch.
Even its competitors seem impressed. OpenAI CEO Sam Altman reportedly sent a memo to employees, according to The Information, in which he acknowledged increasing competition. "We know we have some work to do but we are catching up fast," he reportedly wrote. "I expect the vibes out there to be rough for a bit."
U.S. investing more in nuclear energy for AI
One hallmark of the AI revolution has been its seemingly insatiable need for energy. Tech companies have responded by investing in everything from new green energy sources to experimental satellites to gather the sun's rays to even nuclear reactors back here on Earth. Now, the US government is getting involved too.
The New York Times reports that the Trump Administration is providing a $1 billion loan to help restart Three Mile Island, the nuclear facility in Pennsylvania that, until recently, had been primarily known as the site of the worst nuclear plant accident in the US. Nearly half a century ago, one of the plant's reactors suffered a partial meltdown, triggering a panic that is widely believed to have set back the nuclear industry. Last year, Microsoft announced investments to help restart one of the unaffected reactors, which had laid dormant since it shut down in 2019 due to economic pressures.
Other tech companies have since announced similar nuclear efforts. Earlier this year the Trump Administration said it planned to accelerate new energy projects as part of a series of initiatives to push US "dominance" in AI.
ChatGPT going to school
OpenAI says it's created a version of ChatGPT for K-12 teachers, designed as a "secure workspace to adapt materials for their classrooms, get more out of prep time, collaborate with peers, and get comfortable using AI on their own terms." Schools meanwhile benefit from increased compliance controls. OpenAI said the new tools will be free for verified schools through summer 2027.
The tech giant said the project was a result of its work with organizations such as the American Federation of Teachers. OpenAI also designed a student-focused version of ChatGPT to help with studying and tutoring, announced during the back-to-school season earlier this year. Google too has been expanding its education-focused AI tools, particularly with its document repository technology, Notebook LM as well as through Google Translate.
OpenAI's efforts come as schools continue debating how best to integrate AI into their classrooms. Surveys show that large swaths of both students and teachers are using the tool, and the speed at which it's sweeping through the workforce suggests that future employees will need to be as proficient with it tomorrow as we are today with email and spreadsheets. Still, concerns about cheating and dishonest use have made schools and some students skeptical.
OpenAI acknowledged some of that tension, saying it will work with its first cohort of districts to learn what works, and look for ways to improve. "Together, we’re learning what supports teachers best at scale," OpenAI said in its announcement.
More from MC News
- Run AI Locally: The Best LLMs for 8GB, 16GB, 32GB Memory and Beyond
- Quantization Explained: Why the Same LLM Gives Better Results on High-End Hardware
- Why VRAM and Memory Bandwidth are Key for Powering Local AI
- Hands-on with NVIDIA DGX Spark: Everything You Need to Know
- How to Build a PC with a Hardline Water-Cooling Loop
- 3D Print a Mac Mini Monitor Mount
- The End Has Come for Windows 10: Four Tips to Make the Most of Windows 11
- Everything You Need to Know About WiFi 7
- Keyboard 101: Intro to Computer Keyboards
- Fix It Yourself: Talking to iFixit on Why Repairable Tech Matters
Ian Sherr is a widely published journalist who's covered nearly every major tech company from Apple to Netflix, Facebook, Google, Microsoft, and more for CBS News, The Wall Street Journal, Reuters, and CNET.
