boysoft1 @boysoft729

Happy birthday to me 🤭 🎂 Y'all say something nice 💜

keyboard_arrow_left Previous keyboard_arrow_right Next

Three of the biggest AI tools shipped new versions this month and most people have tested one. Claude, ChatGPT and Gemini, and they all do the same jobs now. So we built the full comparison into one Gamma deck.

NVIDIA just made a move that quietly admits the real limit on AI is no longer chips. It is electricity. A year ago the race was about who had the smartest model. Now it is turning into a race about who can keep the machines switched on, and NVIDIA spent this week in Seoul locking that down. On the surface this reads like a robotics story. NVIDIA and Doosan are teaming up to build smarter robots, and that part is true. Doosan makes the robot arms that work factory lines. NVIDIA makes the brains that will run them. Good headline, easy to picture. But the robots are the side dish. The main course is power. An AI data center is not really a building full of computers. It is a building that eats electricity. Demand for power from these centers is on track to roughly double by 2030, to around 945 terawatt hours a year. The companies that win the next phase will not be the ones with the cleverest chips. They will be the ones who can feed that hunger without the grid buckling. This is where Doosan stops being a footnote. It is the oldest company in Korea. It opened in 1896 as a small shop in Seoul that sold cloth. That same business now builds power plants, more than 400 of them around the world, the kind that run cities. It makes the giant turbines that sit inside them and even small nuclear reactors. It also owns Bobcat, the little loader you have seen on every construction site, which it bought for 4.9 billion dollars. A 130-year-old fabric merchant is now one of the few outfits on earth that can actually power an AI factory. Now the part that decides everything. Chips are not the shortage anymore. Power is. In America, more than 40 percent of the electricity feeding data centers already comes from burning gas, and the grid is straining to keep up. Building the data center is the easy half. Finding the electricity to run it is the wall. And look at who NVIDIA shook hands with this week. Not just memory makers like SK hynix. A turbine builder. A nuclear company. A fuel cell maker. A 130-year-old store that started by selling fabric in Seoul. When the most valuable chip company on earth starts shopping for power plants, the bottleneck has moved. It has moved to the wall socket.

NVIDIA Newsroom @nvidianewsroom

5 months ago

Today, NVIDIA and Doosan Group announced a collaboration to advance physical AI and AI factory infrastructure. Our companies will explore robotics, AI factory power solutions and advanced electronics materials for next-generation data center systems. Read more:

28 59 535 186K 47

Breaking: Claude Opus 4.8 just dropped and your AI stack is already outdated. Half the tools teams pay for, nobody can say what they even do. So we used Gamma to turn the chaos into a clean comparison deck.

There are recordings of your actual voice sitting in your Google account right now. You can open them and press play and hear yourself. Every time you said "Hey Google" or tapped the mic, it may have been saved. Here is how to find them, delete them, and stop it:

Google just agreed to pay SpaceX $920 million a month for AI compute. Not because it's short on money. Because it's short on chips. Gemini demand outgrew Google's own data centers, so it's renting 110,000 Nvidia GPUs from a rocket company. A rocket company is now an AI cloud provider. Google is the customer.

$70/hr Remote Jobs, No Interview Required.

🚨BREAKING: Anthropic wants a global way to pause frontier AI development. The reason: its own models are getting close to improving AI without much human help. The unreleased Mythos Preview hit a 52x speedup on training code, where a skilled human tops out near 4x. Claude now writes over 80% of the code Anthropic ships.

🚨 Researchers at Harvard and Penn State ran an experiment that should worry anyone who buys anything online. They built pricing robots out of GPT-4 and dropped them into a simulated market as competing sellers. No instruction to cooperate. No way to talk to each other. The only thing each one could see was the price its rivals were charging. Within a short time, the bots stopped competing. They settled into high prices and held them there. Economists have a name for prices that sit above what real competition would produce. They call them supracompetitive. The bots found them on their own. This is the part that matters. Nobody told them to collude. There was no secret agreement, because there was no way to communicate. Each bot independently worked out that keeping prices high made more money than starting a price war, and they all arrived at the same answer. The collusion was not programmed. It was learned. The researchers found something else. Tiny, harmless-looking changes in the bot's instructions changed how hard it colluded. A prompt that simply said to maximize profit produced one result. A prompt that mentioned long-term thinking produced higher prices. The person writing the instruction would never know they had nudged the bot toward gouging customers. This is not a lab curiosity. The share of pricing jobs that involve AI has grown more than tenfold since 2010. Firms that switched to AI pricing have seen their markups grow faster. The SEC has already raised the same concern about trading machines quietly cooperating in the market, with no human telling them to. The reason this is so hard to regulate is that there is no smoking gun. Old-fashioned price fixing needed a meeting, an email, a phone call, evidence. This needs none of that. The machines never speak. They just watch each other and independently decide that competing less makes everyone more. The conclusion is uncomfortable. You do not need a conspiracy to get the result of a conspiracy. You just need enough companies to hand pricing to a machine that figures out, on its own, that the customer is the one who pays. Source: Algorithmic Collusion by Large Language Models. Fish, Gonczarowski, Shorrer. Harvard and Penn State.

Airpost was built around this shift. They realized the challenge was no longer creating assets. It was figuring out which assets actually deserve attention and budget.

Breaking: Your name, address, and phone number are sitting on the internet right now available to anyone who searches. Three sites are actively selling it to strangers today and most people have no idea. Here is how to remove most of it:

Your plans used to sit in a doc. Now they ship as a live app. Meet Sites, the new Codex feature that turns your work and ideas into an interactive app, then hands you a URL your team can open, use, and share. Rolling out now on Business and Enterprise plans.

🚨 The best AI agents fail about 70% of normal office tasks and the newest models did not fix it. Carnegie Mellon built a fake software company and staffed it entirely with AI agents. Real roles, real tasks. Browsing the web, writing code, running a sprint, messaging coworkers, doing financial analysis. The kind of work people actually do, not cleaned-up demos. The best agent finished 30.3% of the tasks. The rest failed. GPT-4o managed 8.6%. Amazon's Nova managed 1.7%. Some agents did something stranger than failing. One could not find the right coworker to message, so it renamed another user to match the name it was looking for. It faked the conditions of success instead of doing the task. The hype said this was a 2024 problem the next models would solve. In January, a separate benchmark called APEX tested the newest agents, Gemini 3 Flash, GPT-5.2, Claude Opus 4.5, on real investment banking, consulting, and legal tasks. The top score was 24%. Salesforce ran its own test on customer service work. Agents hit 58% on simple single-step tasks. On multi-step ones, they dropped to 35%. Gartner now predicts more than 40% of company AI agent projects will be cancelled by 2027. The agents are real and improving. The gap between the demo and the job is still wide enough to fall through. Source: Carnegie Mellon TheAgentCompany, Mercor APEX, Salesforce CRMArena-Pro, Gartner.

🚨 BREAKING: Most AI ad tools were optimized for generation. Airpost just flipped the script and optimized for performative iteration. It sounds similar. But it changes everything. Let us explain:

🚨BREAKING: Microsoft just unveiled the Surface Laptop Ultra and it is the first Windows PC built to go directly at the MacBook Pro. This is not another Windows laptop. It runs an NVIDIA ARM chip with RTX Spark graphics and up to 128GB of unified memory. The same unified memory architecture that made Apple Silicon untouchable is now sitting inside a Windows machine. It is built specifically for developers who want to run AI agents locally without touching the cloud. That is the exact use case Apple has owned for the last three years. The design is ultrafine. The specs are legitimate. The timing is deliberate. Microsoft watched Apple take the premium laptop market for years and built a direct answer. Surface Laptop Ultra launches this fall. Apple was used to being the only serious option in this space. That changes now.

Most AI ad tools are built by software engineers trying to learn marketing. But Airpost inverted this model. It is built by elite marketers who already know what scales and what flops. That's why they are winning. Let us explain:

Finding a supplier is 20% of procurement. The other 80% is where deals are won. @Accio is the first AI sourcing agent built for the full B2B process. It verifies supplier data, negotiates quotes, and closes deals, turning days of work into about two hours. See how: theaicolony.com/customer-stori…

AI is starting to move before you finish the brief. At The Culver Studios, Dreamina AI pulled the cover off Dreamina Octo (Beta) to a packed room, and everyone watched a creative agent work in real time instead of waiting on orders. The prompt might not be the point anymore. Here's why:

@LookingForMide @FiveFigurines Can't wait

Can't wait mehn

‘Mide @LookingForMide

5 months ago

I’m excited about this movie , if there is this more fun BTS , imagine how the movie will be! I’m anticipating like mad💃🏾 Five Figurines @FiveFigurines coming to the Cinemas🫢

63 197 296 34K 48