
Nano Banana is such a hit that Gemini is suddenly ahead of ChatGPT, at least if we’re measuring by the app store. Lots of interesting new data about how people are actually using AI. A new accelerator program from OpenAI itself, and is the AI boom like the dawn of the microprocessor or more like the dawn of containerization?
Loading summary
A
Welcome to the Tech Brew Ride home for Monday, September 15, 2025. I'm Brian McCullough. Today, Nano Banana is such a hit that Gemini is Suddenly ahead of ChatGPT. At least if we're measuring by the App Store. Lots of interesting new data about how people are actually using AI. A new accelerator program from OpenAI itself. And is the AI boom like the dawn of the microprocessor or more like the dawn of containerization? Here's what you missed today in the world of tech. Do you know what the number one free app is in the App Store right now? It's Google's Gemini app. Gemini is not Only now beating ChatGPT, it is beating everything. Apparently a ton of you are using the heck out of that Nano Banana image model, which has been used more than 500 million times apparently to make images. Look out Adobe. But as Chad Hurley tweeted, incredible growth without goonwear or rage baiting. Who knew mind blown emoji. And as Kimon Isthmus tweeted, nanobanana has done Google a huge favor. It is a tool suitable for mass use that is fun to use and that contributes greatly to the spread of Gemini. Google has done itself a huge favor with this quoting 9 to 5 Google from the launch on August 26 to September 9, the Gemini app saw 23 million new users. Nano Banana has been used to edit over 500 million images in that period. This image edit editing model is a viral hit due to how it maintains character, likeness and consistency. You can upload multiple photos to make new ones, transfer styles, and of course conversational editing. All free users can generate or edit up to 100 images per day. While that goes to 1000 if you're a subscriber at $19.99 plus per month, end quote. And quoting Tom's Guide, the feature has gone viral on social platforms. Because it's accessible, you don't need expert prompting to get good results. It's visual, and turning a selfie or sketch into a figurine is instantly shareable. It's also fast. The model runs quickly enough that edits feel playful, not tedious. It's a reminder that one well designed feature can completely change the trajectory of an app. Actually, we've got a whole bunch of data all of a sudden about how y' all are using these AI tools because some of these AI companies are telling us how. For the for instance, OpenAI has released the first detailed public study on how people use ChatGPT and according to them, 73% of chats were non work related. Practical guidance was the top use case and more. Quoting the Washington Post, the company reports that most ChatGPT users are women and that the majority of requests sent its way are not work related. The user base is dominated by young people. Nearly half of the conversations studied were from people aged 18 to 25. OpenAI released the data Monday in a 62 page research paper, which has not been peer reviewed. It is based on Chat logs from 1.5 million of ChatGPT's users between May 2024 and June of 2025, when ChatGPT was first gaining traction. In late 2022 and early 2023, around 80% of users had typically masculine first names, the company's study found. By June 2025 that had flipped 52% of ChatGPT users have feminine names, OpenAI researchers wrote. ChatGPT usage is growing faster in poorer countries than it is in wealthy ones, the study said. Although it did not list usage by country, OpenAI has said ChatGPT now has more than 700 million weekly users. People are increasingly using ChatGPT in their personal lives rather than for help at work, OpenAI's study found. In June 2024, prompts to the chatbot were split roughly evenly between work and personal uses. By June of 2025, non work uses made up 73% of all conversations, the company said. OpenAI sorted the more than 1 million chats studied across that period into seven categories. The biggest was practical guidance at 28.3% of all chats, a category the company's researchers defined as including people seeking how to advice, help with schoolwork and tips on, say, working out. The second biggest category was people asking ChatGPT for writing help, a use case that builds on the advances in a text generation that birthed the chatbot and others like it. OpenAI's study found that the most popular writing task given to ChatGPT was editing or critiquing text, followed by personal writing or communication, a category defined to include helping with emails or social media posts. The ability of ChatGPT and similar AI tools to generate computer code has been enthusiastically adopted by many in the tech industry. OpenAI's study found that around 4.2% of the sample chats were related to programming. The third most popular use case for ChatGPT across the data includes included in the study was seeking information, a category the researchers said appears to be a very close substitute for web search. End quote. Yeah, but let's come back to that coding use case because Anthropic has their own report out detailing how CLAUDE usage varies by country and US states, but they found 36% of folks use it for coding, 77% of enterprise use is for automation and more, quoting Bloomberg More than 3/4 77% of companies Usage of Anthropic's Cloud AI software involved automation patterns, often including full task delegation, according to a research report the startup released on Monday. The finding was based on an analysis of traffic from Anthropic's application programming interface, which is used by developers and businesses. On the whole, Anthropic found businesses primarily use CLAUDE for administrative tasks and coding, the latter of which has been a key focus for the company and much of the AI industry. Anthropic, OpenAI and other AI developers have released more sophisticated AI tools that can write and debug code on a user's behalf. Anthropic's Chief Executive Officer Dario Amodai has previously issued particularly dire predictions, claiming AI could wipe out 50% of entry level jobs and that people should stop sugarcoating what lies ahead. More from Anthropics Report Quote Education and science usage shares are on the rise while the use of CLAUDE for coding continues to dominate our total sample at 36%. Educational tasks surged from 9.3% to 12.4% and scientific tasks from 6.3 to 7.2%. Users are entrusting Claude with more autonomy. Directive conversations where users delegate complete tasks to Claude jumped from 27% to 39%. We see increased program creation encoding which was up by 4.5 percentage points, and a reduction in debugging which was down by 2.9 percentage points, suggesting that users might be able to achieve more of their goals in a single exchange. While computer and mathematical tasks still dominate overall usage at 30. 36%, we are seeing sustained growth in knowledge intensive fields, end quote. Not sure what sort of tea leaves to read here, but China says Nvidia has been found in violation of antitrust regulations in a preliminary investigation In a brief statement without further details quoting Bloomberg, the US Chipmaker was found in violation of antitrust regulations after the acquisition of networking gear maker Mellanox Technologies, the state Administration for Market Regulation said after concluding a preliminary investigation. The surprise announcement emerged with US And Chinese officials heading into a second day of wide ranging negotiations in Madrid over tariffs which could shape the relationship between the world's two largest economies. Over the weekend, China also said it was launching an anti dumping investigation targeting a type of semiconductor made by US companies including Texas Instruments. The regulator didn't specify on Monday what sort of remedies it would seek from Nvidia, adding it will investigate the company immediately. Respond to an email request for comment outside of regular office hours on Monday, President Donald Trump said he would speak with China's Xi Jinping on Friday after lower level negotiations between the two countries. U.S. treasury Secretary Scott Besant said the two sides had discussed the poor timing of the Nvidia investigation, but had agreed on a framework to keep ByteDance's TikTok app running in the US another long running conflict between Washington and Beijing. End quote.
B
Eczema isn't always obvious, but it's real and so is the relief from Ebglis. After an initial dosing phase, about 4 in 10 people taking EBGLIS achieved itch relief and clear or almost clear skin at 16 weeks, and most of those people maintain skin that's still more clear at one year with monthly dosing.
C
Eglis Librekizumab LBKZ a 250mg per 2ml injection, is a prescription medicine used to treat adults and children 12 years of age and older who weigh at least 88 pounds or 40 kilograms with moderate to severe eczema, also called atopic dermatitis that is not well controlled with prescription therapies used on the skin or topicals, or who cannot use topical therapies. EBGLIS can be used with or without topical corticosteroids. Don't use if you are allergic to ebglis. Allergic reactions can occur that can be severe. Eye problems can occur. Tell your doctor if you have new or worsening eye problems. You should not receive a live vaccine when treated with Epglis. Before starting Epglis, tell your doctor if you have a parasitic infection.
B
Searching for real relief? Ask your doctor about EBGLIS and visit epgliss.lilly.com or call 1-800-LilyRx or 1-800-545-5979.
A
We've all been there. Too many SaaS tools, not enough visibility at all, and way too much access for you to keep track of. It's the stuff security nightmares are made of. That's where Trelica by1Password comes in. They inventory every app your company uses and create app profiles to help you easily assess risks, manage access, and make sure your password security is locked down tight. With 1Password Passwords Extended Access Management, you can control your company's many, many SaaS tools securely onboard and offboard your people and actually hit your compliance goals. I've been telling you about 1Password Extended Access Management all year and now Trelica comes along to make things even better. Sleep Easy with Trelica by 1Password Learn more at 1Password.com Ride that's 1Password.com Heard of Klarna? You probably have. They burst onto the e commerce scene back in 2005 to challenge traditional credit during a time when consumer trust was at an all time low. Since then, Klarna has won the trust of 111 million customers by providing a more flexible, transparent way to shop and pay online. And now they're also listed on the New York Stock Exchange, underscoring their scale and credibility. In case you didn't know, Klarna helps consumers split the cost into four interest free payments with no hidden fees, no confusing terms and no revolving debt. Klarna has helped save consumers billions in interest. Millennials and Gen Z are also using this AI powered payment tool to help budget smarter and avoid interest fees. Learn how you can pay smarter with klarna@l.klarna.com 22xc KlarnaTechbrew and look out for the pink Klarna logo at checkout next time you shop online. That's l.klarna.com PayPal has launched PayPal links, which lets users create a personalized one time link to send or request payment to other PayPal users. Quoting ZDNet Move over Zelle, watch out. Cash app PayPal has unveiled a new type of peer to peer payment method that promises to be easier and simpler than other options. On Monday, the online payment service unveiled PayPal links, a new way for individuals to send and receive money without incurring fees or or tax forms. The new option is now available in the US to start and will expand to the uk, Italy and other countries later in September. Whether you want to pay someone or request payment, PayPal links lets you generate a personalized one time link that you can share via a text, direct message or email. To do that, just launch the PayPal app for iOS or Android. Select the PayPal link option, add the amount you want to pay or request, and then copy the link into your message or other conversation. You can also add a note or even an emoji. After tapping the link, the person on the other end either sends you your requested payment or accepts the money you sent. The receiver then gets immediate access to the funds. Since the transaction runs through the PayPal app, both the payer and payee need a PayPal account. If a link is shared with someone who doesn't have an account, that person is prompted to create one through the PayPal app or website to safeguard your security and privacy. Each link is private, designed for one time use and generated just for a specific transaction. Unclaimed links expire after 10 days. Before the time is up, you can send the other person a reminder or even cancel the payment. With PayPal links, PayPal collects no transaction fee. Payers and payees are also exempt from 1099k reporting, so you won't receive any tax forms for the payment. Finally, PayPal has one more surprise Cryptocurrency will be included in the links feature. People in the US will soon be able to quickly send Bitcoin, Ethereum, PiusD and other cryptocurrencies into PayPal, Venmo and other supported digital wallets. End quote and founders, I'm putting this on your radar. OpenAI has unveiled Grove, a five week mentorship program for nascent tech entrepreneurs hosted in its San Francisco headquarters with around 15 participants in its first cohort. Quoting CNBC, OpenAI on Friday introduced a new program dubbed the OpenAI Grove for early tech entrepreneurs looking to build with artificial intelligence and applications are already open. Unlike OpenAI's Pioneer program, which launched in April, Grove is aimed toward individuals at the very nascent phases of their company development from the pre Idea to Pre seed stage. For five weeks, participants will receive mentoring from OpenAI technical leaders, early access to new tools and models, and in person workshops located in the company's San Francisco headquarters. Roughly 15 members will join Grove's first cohort, which will run from October 20 to November 2021. Applicants will have until September 24 to submit an entry form. CNBC has reached out to OpenAI for comment on the program. Following the program, Grove participants will be able to continue working internally with the ChatGPT maker. Finally today, a really interesting weekend read from Jerry Newman over at Colossus. The gist of his argument is that generative AI is a real technological shift, but the way it will create wealth will likely disappoint most investors. Chasing the next big thing Neumann argues that AI today looks less like the microprocessor, an innovation that spawned thousands of startups and fortunes, and more like containerized shipping, a society wide transformation whose gains largely flowed to customers and incumbents, not to waves of new independent winners. The right investing question isn't how big is AI? It's who actually captures the value. In classic tech cycles you get phases like early interruption or experimentation, frenzy and speculation, synergy and productization and maturity commoditization. Investors tend to do well during frenzy and synergy, but in late Wave moments. Competition is immediate. Profits get competed away quickly. Neumann argues. AI fits that late wave pattern. It arrived with minimal surprise, sits atop established computing data and cloud foundations and immediately triggered a free for all rather than a long open ended tinkerer phase. That dynamic caps excess returns for most builders. Follow the money and the picture clarifies. If you didn't get into a frontier model company early, you probably missed the only part of the stack where outsized gains could accrue high capex brutal competition and eventual consolidation. Validation mean only a handful of model providers funded and controlled by tech giants are likely to dominate domain specific models. For example for legal or coding may be valuable, but will likely be rolled up. App layer startups that rely on those models will see their economies squeezed by platform pricing or vertical integration. Success at the app layer can perversely invite capture by the model owners. Expect convergence toward AI companies few in number rather than a broad enduring independent app ecosystem. Could intermediaries fare better? Maybe firms that broker between enterprises and model providers or protect proprietary data think picks and shovels like data governance or interface layers. But the shipping analogy suggests even successful intermediaries top out at mid size outcomes because dominant platforms won't allow them strategic leverage. What about moving upstream into chips, data centers and specialized data? Some value will accrue there, but much of it is already priced in and capex cycles can turn vicious if AI spending plateaus or dips. Suppliers stuck with long lead commitments can be caught with excess capacity. General data is commoditized only timely, hard to replicate. Domain specific data sets hold pricing power and even that market isn't huge. The downstream story is more promising. AI is particularly good at extracting high quality output from ambiguous inputs. That raises productivity in knowledge intensive sectors professional services, healthcare, education, finance and creative work which together represent a third to half of global gdp. Companies whose strategies already assume lower costs think Ikea's quality at low price make it up in volume Playbook are primed to benefit fastest but simply banking cost savings as margin rather than reinvesting to grow is a losing strategy over time. Macro wise economists estimate AI could lift global GDP by 1 to 7% over the next decade. Trillions of dollars in output, but most of that value likely accrues to consumers via che better knowledge services while human intensive face to face services become relatively more expensive. Investors should therefore fish downstream back businesses that can convert knowledge worker productivity into new demand products and markets. In this era the edge is not betting on AI itself, but betting on the opportunities AI unlocks. As I said it was a very interesting read. Click through to read the whole thing. Quick note that the show tomorrow will be like an hour or so late. I've got something scheduled in the morning that I need to bang out, but also important news for all listeners. As you know, we recently joined the Morning Brew family and as I said, some of that was in aid of hopefully getting a different level of sponsors. So help me do that. We just got started with Morning Brew, so they need to know more about who is listening to this show. In the eight years of this show, I've never done a listener survey. And so the bottom link in the show notes today is to a survey asking you to tell us about yourself, about your listening habits. Two reasons I hope you'll help me out. Number one, I put some questions in there asking what it is you want to hear more of and what you want to hear less of. So this is designed to get demo information on the audience, but also for you to help guide how the show evolves editorially. Take the time and tell me what you like and what you don't like.
B
Like.
A
And for your time and your trouble, you'll be entered into a raffle to win a $500Amex gift card. That's right. One of you will win a $500 gift card if you take the survey. So thanks in advance for doing that again. Link in the bottom of the show notes to the survey. Help me make the show better. Talk to you tomorrow.
B
Trip Planner by Expedia.
A
You are made to outdo your holiday.
B
Your hammocking and your pooling. We were made to help organize the competition. Expedia made to travel.
Episode: Gemini’s Suddenly The #1 App In All The Land
Host: Brian McCullough
Date: September 15, 2025
In this episode, Brian McCullough explores the rapid rise of Google's Gemini app to the top of the App Store, primarily powered by the viral Nano Banana image model. He reviews fresh usage data for leading AI tools like ChatGPT and Claude, shines a spotlight on OpenAI’s new startup accelerator, touches on a breaking antitrust investigation into Nvidia, and ends with a commentary on how the AI boom compares to previous technological revolutions. The episode provides insight into how everyday people and enterprises are actually using AI, and what the real winners and losers of the AI era are likely to be.
[00:04 - 03:30]
Gemini App Hits #1:
“Do you know what the number one free app is in the App Store right now? It's Google's Gemini app. Gemini is not only now beating ChatGPT, it is beating everything.” — Brian McCullough [00:22]
Nano Banana Model’s Role:
User Adoption & Virality:
Social Media Impact:
Community Reactions:
[03:30 - 08:37]
New Study Released: From 1.5 million ChatGPT users (May 2024 – June 2025).
Key Findings:
Specific Use Cases:
“People are increasingly using ChatGPT in their personal lives rather than for help at work… non work uses made up 73% of all conversations.” — Brian McCullough [04:33]
Usage by Business:
“Directive conversations where users delegate complete tasks to Claude jumped from 27% to 39%. We see increased program creation in coding … suggesting users might be able to achieve more of their goals in a single exchange.” — Brian McCullough quoting Anthropic report [06:47]
Anthropic CEO Dario Amodei:
[07:40 - 08:37]
Nvidia Faces Chinese Antitrust Probe:
“The surprise announcement emerged with US and Chinese officials heading into a second day of wide ranging negotiations in Madrid over tariffs…” — Brian McCullough [07:59]
[09:38 - 11:00]
PayPal Links:
“PayPal collects no transaction fee. Payers and payees are also exempt from 1099k reporting, so you won’t receive any tax forms for the payment.” — Brian McCullough [10:45]
[11:00 - 12:00]
OpenAI Grove:
“Unlike OpenAI’s Pioneer program ... Grove is aimed toward individuals at the very nascent phases of their company development ... applications are already open.” — Brian McCullough [11:12]
[12:00 - 15:30]
Jerry Neumann’s Analysis:
“If you didn’t get into a frontier model company early, you probably missed the only part of the stack where outsized gains could accrue ... expect convergence toward AI companies few in number rather than a broad enduring independent app ecosystem.” — Brian McCullough summing up Neumann [13:50]
On Gemini’s Rise:
“A ton of you are using the heck out of that Nano Banana image model … Look out Adobe.” — Brian McCullough [00:19]
On Simple, Viral AI Features:
“It’s a reminder that one well designed feature can completely change the trajectory of an app.” — Tom’s Guide via Brian McCullough [02:20]
On ChatGPT’s Changing Demographics:
“By June 2025 that had flipped 52% of ChatGPT users have feminine names, OpenAI researchers wrote.” [04:04]
On the Limits of the AI Boom:
“The right investing question isn’t how big is AI? It’s who actually captures the value.” — Brian McCullough summarizing Jerry Neumann [13:20]
This episode delivers rapid-fire yet deep insight into how AI is permeating not just the tech industry but everyday life and business. With Gemini’s explosion in popularity and tangible data on how AI is used, listeners are prompted to rethink assumptions about who actually benefits from these shifts. The take-home: The real story in AI isn’t just elite models or startups, but how practical, viral user features—and the right business strategies—will shift value and power in the years ahead.