Here’s some more brief news from today, with a few unfiltered comments and opinions that are often missing in this world of news.
Got your popcorn ready? Today’s barrage of updates would have filled an entire quarter just two years ago. Big Tech is now playing for the future with daily moves in AI and open models, and the competition is getting tighter every day.
#Meta is celebrating one billion downloads of Llama, its open-source model. A milestone that really makes you think—the developer community is clearly rewarding the open approach. One billion is no small number, and Meta is showing everyone that the future is won through sharing, not secrecy.
…and #Google has also launched TxGemma, new open AI models designed to revolutionize drug discovery. It’s a bold strategic move: Google knows full well that AI can radically change the pharmaceutical industry. In fact, it already has. We’re now in the hype phase of AI for pharma—let’s hope that beyond all the grand promises of revolutionary breakthroughs, there’s real substance.
Big G has added new features to #Gemini, including Audio Overviews. It’s definitely interesting, but how many more tools will it take before someone actually simplifies our work instead of complicating it with a hundred functions no one will ever use? Maybe we need to go back to basics instead of piling on more complexity.
And then, Boom! #Nvidia kicks off its event with the legendary Jensen Huang. Two hours of keynote to announce chips, spectacular robots, and autonomous vehicles. Nvidia makes us dream, but companies’ investments in AI need to quickly translate into real, accessible solutions—or else we risk being mesmerized by a beautiful but empty show.
#Adobe has long been in the game—first with AI in Photoshop and now launching its new system of specialized AI agents for businesses. Ten agents that promise to revolutionize marketing, customer experience, and business workflows. The idea is great, but are we sure they won’t just become another excuse to drive up licensing fees without adding tangible value? We hope and wait, as always, for proof on the ground.
Also, Liam Fedus, VP of Research at #OpenAI, is about to launch a startup specializing in advanced materials with AI. It’s interesting that OpenAI itself has decided to fund and collaborate directly with it. Good for the sector, but bad for companies that are left behind, finding themselves without any solid pillars.
Meanwhile, one startup in advanced materials reminds us that AI isn’t just about chatbots and automatically generated texts—it’s making a real impact, with results you can physically feel. Even in the real world. This is where we’re likely to see the most genuine innovations.
Speaking of genuine innovations, #StabilityAI has unveiled a new model that promises to create dynamic 3D videos from a single image. Extremely useful for the 3D projects I work on—which used to take days on Blender. And #Tencent, with Hunyuan, is diving into high-quality 3D model generation with two new specialized products. But how usable will these models really be in real life? So far, I’ve only seen tools that generate output requiring more time to fix than to recreate from scratch on Blender.
On the technical and practical side, #Unstructured promises to beat Gemini 2.0 at document management thanks to what they call “contextual chunking.” It sounds like a classic David versus Goliath challenge—small but effective against big and powerful. The fact remains: documents and data are becoming increasingly crucial in the AI world, and whoever manages them better and sooner will capture a significant slice of the market.
In short, the game is wide open and the major players aren’t holding back with daily hits, announcements, and strategies.
Done with your popcorn?
