- Luis Poveda's AI Newsletter
- Posts
- Luis Poveda's AI Newsletter: March 03, 2025
Luis Poveda's AI Newsletter: March 03, 2025
GPT-4.5, Sora, Claude 3.7 Sonnet, and more.
Luis Poveda
March 03, 2025
Executive Summary
This week has seen major developments in the AI landscape with multiple frontier models being released. Anthropic launched Claude 3.7 Sonnet with breakthrough reasoning capabilities, while OpenAI unveiled GPT-4.5 "Orion" model. Consumer AI also advanced with Amazon's new Alexa+ generative assistant, and OpenAI's Sora video generator finally became available in Europe. Meanwhile, hardware constraints continue to challenge the industry, with OpenAI's CEO highlighting critical GPU shortages.
AI Model Launches and Updates
OpenAI Unveils GPT-4.5 "Orion" as Largest Model to Date
OpenAI has released GPT-4.5, codenamed "Orion," its largest and most compute-intensive model to date. The new model is initially available only to $200/month ChatGPT Pro subscribers, with plans to roll out to Plus and Team users next week. While OpenAI hasn't disclosed the model's specific size, they claim it delivers more nuanced understanding of human emotions and interactions, with reduced hallucinations due to its expanded knowledge base.
"If you know more things, you don't need to make things up".
OpenAI's Sora Video Generator Now Available in Europe
OpenAI has finally expanded access to its Sora video generation model to users in the European Union, United Kingdom, Switzerland, Norway, Liechtenstein, and Iceland. The powerful text-to-video generator, which was initially released to US users in December 2024, is now available to ChatGPT Plus and Pro subscribers in these European regions. This expansion comes after OpenAI addressed regulatory compliance requirements for the European market.
"Sora has arrived in the EU and the UK".
Anthropic Launches Claude 3.7 Sonnet with Revolutionary "Thinking" Capabilities
Anthropic has released Claude 3.7 Sonnet, billing it as the industry's first "hybrid AI reasoning model." This groundbreaking model combines real-time responses with an extended "thinking" mode that allows it to reason through complex problems for as long as users want. The model is rolling out to all users, but only those on premium plans will access the advanced reasoning features.
"Similar to how humans don't have two separate brains for questions that can be answered immediately versus those that require thought, we regard reasoning as simply one of the capabilities a frontier model should have, to be smoothly integrated with other capabilities, rather than something to be provided in a separate model".
AI Industry Challenges
OpenAI CEO Sam Altman Warns of Critical GPU Shortage
OpenAI CEO Sam Altman has revealed that the company is "out of GPUs," forcing a staggered rollout of their new GPT-4.5 model. The GPU shortage is significantly impacting the company's ability to scale, with Altman stating that "tens of thousands" more GPUs will be needed before additional ChatGPT users can access the new model. This hardware constraint highlights the growing challenges facing AI companies as they develop increasingly compute-intensive models.
"We've been growing a lot and are out of GPUs. We will add tens of thousands of GPUs next week and roll it out to the Plus tier then... This isn't how we want to operate, but it's hard to perfectly predict growth surges that lead to GPU shortages".
AI Events and Forecasts
Mobile World Congress 2025: AI Integration Expected to Dominate
The Mobile World Congress 2025, set to begin on March 3 in Barcelona, is expected to showcase significant advancements in AI integration for mobile devices. Industry analysts anticipate that on-device AI capabilities will be the central theme, with major smartphone manufacturers unveiling new AI processors and features. Among the highlights, Realme will launch its 14 Pro series featuring AI-optimized features and innovative color-changing technology.
"Last year, the show hosted over 2,700 exhibitors and 101,000+ attendees. The GSMA, which holds the event, says 2025 will be even bigger. In addition to fresh hardware, we'll see the latest developments in the software that powers them. Translation: Expect lots of AI".
Robotics and Automation
Figure AI Drops OpenAI Partnership and Secures Massive New Funding
Figure AI, a promising humanoid robotics startup, has announced it's exiting its partnership with OpenAI in favor of developing in-house AI models for its robots. Concurrently, the company is reportedly in talks to raise $1.5 billion in a Series C round that would value the company at a staggering $39.5 billion, with Parkway Venture Capital expected to lead the round. This development highlights the growing investor confidence in commercial applications for advanced humanoid robots.
"Today, key vendors — xAI, Meta, IBM, Boston Dynamics, Agility Robotics, Apptronik, Figure.ai, FourierIntelligence, and Sanctuary.ai — have [humanoid robots] inching their way toward the workforce".
Conclusion
The AI landscape continues to evolve at breakneck speed, with companies prioritizing two distinct paths forward. On one hand, we're seeing increasingly powerful and computation-heavy frontier models like GPT-4.5 and Claude 3.7 Sonnet that push the boundaries of reasoning and understanding. On the other hand, we're witnessing the practical integration of AI into everyday consumer technologies, from Amazon's Alexa+ to AI-optimized smartphones. The critical hardware bottleneck highlighted by OpenAI's GPU shortage represents a significant challenge for the industry's ambitions, potentially slowing the pace of innovation until infrastructure can catch up with software capabilities. As we look ahead, the commercialization of humanoid robotics adds yet another dimension to AI's expanding reach, suggesting we're on the cusp of seeing AI move beyond screens and voice assistants into physical, autonomous entities that can interact with and manipulate the world.
The Author
 Luis Poveda’s AI Newsletter | Luis Poveda is a technology optimist and passionate innovator, constantly exploring and researching the latest trends. Based in Barcelona, he is currently focused on AI and developing a modern AI-driven network observability tool. He is also the creator and maintainer of Luis Poveda's AI Newsletter, where he curates and shares key insights on the evolving AI landscape. |