Modelos, agentes, investigación y negocio de la IA.
Tesla has filed a trademark for 'Megapod' for a modular AI data center hardware system. The product would include servers, networking, power distribution, and cooling in a single module, challenging Nvidia in the AI infrastructure market.
Google DeepMind has announced a $75 million investment in A24, the independent studio behind hits like 'Everything Everywhere All at Once'. It is the first time an AI lab has taken a stake in a Hollywood studio, opening a new chapter in the relationship between artificial intelligence and cinema.
The cybersecurity industry faces a dilemma: bet on platform consolidation or a hybrid AI approach combining specialized models. We analyze the implications for businesses, users, and the market.
Multi-agent pipelines are gaining ground over single agents in AI applications. A recent article from Towards Data Science explains how this approach improves accuracy, control, and scalability in complex tasks like text-to-SQL. We analyze why it matters and what it means for the future of AI development.
OpenAI and Broadcom have launched Jalapeño, an ASIC chip designed exclusively for inference of large language models. The announcement marks a milestone in OpenAI's custom hardware strategy and could redefine the economics of AI.
Qualcomm has acquired Modular, a promising AI chip software startup, for nearly $4 billion. The deal aims to strengthen its position in on-device AI, directly competing with NVIDIA and other tech giants.
The US government has asked Meta to share its artificial intelligence models for review, amid a global debate on AI risks. This move could set a significant regulatory precedent.
Faced with the shift toward usage-based billing for AI models, database vendors present themselves as a solution to control costs. Pinecone launches Nexus, a 'knowledge engine' that prevents agents from rediscovering data each time, saving tokens and improving efficiency.
The rise of AI and new regulations are forcing companies to rethink how they manage their data. Centralization is no longer viable: movement costs, leakage risks, and sovereignty laws demand a distributed approach.
Z.ai has released GLM-5.2, an open-source model with MIT license that surpasses GPT-5.5 in programming benchmarks. However, its hundreds of billions of parameters require datacenter infrastructure, posing a dilemma between performance and accessibility.
Legion LegalTech has filed a lawsuit against the U.S. government over the directive that forced Anthropic to deactivate its most powerful AI models globally. The company alleges the shutdown has caused 'immediate, irreparable, and existential' harm to its business, which relied on those models for legal analysis services.
During a government test, Anthropic's Mythos model identified vulnerabilities in US classified systems within hours, marking a milestone in AI cybersecurity.