Google DeepMind and Google Research launched Weather Lab, an interactive platform featuring an AI-driven cyclone forecasting model that simulates up to 50 storm scenarios as much as 15 days in advance. Initial collaborations with the U.S. National Hurricane Center show the AI model predicts cyclone tracks an average of 87 miles closer to actual landfall points than leading European physics-based methods. Trained on decades of ERA5 reanalysis data, the stochastic neural-network model excels at capturing path, intensity, size, and shape of tropical cyclones, all within minutes. While promising, the tool is intended for research support rather than operational reliance, with forecasters advised to integrate AI outputs alongside traditional models. Google emphasizes ongoing improvements and partnerships with global meteorological agencies to refine performance. The innovation points to a future where AI augments disaster-response workflows, potentially saving lives by providing earlier and more accurate warnings.

OpenAI unveiled o3-pro, its latest reasoning-focused model, positioning it as a more affordable alternative to o3 for Pro and Team subscribers. The upgrade emphasizes improved capabilities across scientific research, code generation, and business analytics, addressing customer demands for deeper, more reliable reasoning. By slashing prices by 80 percent, OpenAI aims to accelerate adoption among mid-market enterprises and developers who previously found premium models cost-prohibitive. Early benchmarks show o3-pro outperforming its predecessor on complex logic tasks and multi-paragraph comprehension tests. OpenAI’s move follows competitive pressure from Anthropic’s Claude 3 and Google’s Gemini, both of which have recently advanced their own reasoning architectures. This aggressive pricing strategy signals a shift toward volume-driven AI offerings, where lower margins may be offset by increased usage and platform stickiness. Analysts predict that o3-pro could become the workhorse model for companies embedding AI into day-to-day workflows, from automated report drafting to real-time data analysis.

Google, Scale AI’s largest customer, plans to sever its partnership following Meta’s $14.8 billion acquisition of a 49 percent stake in the data-labeling specialist. According to insiders, Google fears proprietary training data could indirectly benefit its direct competitor. The decision reflects broader shifts in the data-infrastructure market, where control over high-quality labeled data has become a strategic asset. Scale AI’s annotation services underpin many leading AI models, and Meta’s deal grants it significant influence over a key supplier. Clients such as Tesla and OpenAI are now re-evaluating their annotation pipelines, exploring alternatives like Hive and Alegion. The split could spark a wave of new partnerships and consolidation in the annotation space. While Google explores in-house labeling solutions, Meta’s move cements its ambition to control both model development and the data that trains those models. The fallout underscores how critical data partnerships have become in competitive positioning among hyperscalers and social-media giants.
Source: reuters.com

Meta Platforms is acquiring a 49% stake in Scale AI for approximately $14.8 billion, according to Reuters. Scale AI specializes in data labeling and model evaluation for clients like OpenAI and the US government. It generated around $870 million in 2024 and is projected to exceed $2 billion in 2025. As part of the deal, Scale’s CEO Alexandr Wang will join Meta to oversee a new "superintelligence" lab. This partnership aims to secure a steady supply of high-quality training data, address Meta’s internal AI setbacks (like Llama 4 and delayed Behemoth), and strengthen its position in foundational model development. The structure appears designed to avoid antitrust scrutiny, marking a major strategic move by Meta to vertically integrate data resources reuters.com.

OpenAI has entered a strategic cloud partnership with Google despite its rival status, according to multiple Reuters reports. With soaring demand for ChatGPT and other AI services, OpenAI is supplementing Microsoft Azure with Google Cloud's TPUs via CoreWeave. This collaboration aims to prevent infrastructure bottlenecks and reduce reliance on a single provider. It highlights how compute needs are reshaping competitive dynamics and cloud alliances within the AI industry.