Etiquetas: compute infrastructure

Elon Musk’s artificial‑intelligence unit xAI posted a $6.4 billion operating loss on $3.2 billion of revenue for 2025, according to SpaceX’s recent IPO filing. The filing also reveals plans to scale the Grok model to “multiple trillions of parameters,” a move that will drive even higher compute costs. While the combined SpaceX‑xAI enterprise eyes a potential $1.75 trillion market debut, user adoption of Grok remains modest, with only about one‑fifth of the 550 million monthly active users across the ecosystem engaging the AI features. Leer más

AI video‑generation startup Runway announced a $315 million Series E financing round that lifted its valuation to $5.3 billion. The new capital will fund the pre‑training of next‑generation world models and expand the company’s research, engineering, and go‑to‑market teams. Runway’s latest model, Gen 4.5, delivers high‑definition video from text prompts with native audio, long‑form generation, character consistency, and advanced editing tools, outperforming competing offerings from Google and OpenAI on key benchmarks. The company also broadened its compute capacity through a partnership with CoreWeave and is seeing growing interest from gaming and robotics sectors. Leer más

OpenAI announced a multi-year agreement with AI chipmaker Cerebras to deliver 750 megawatts of compute power from this year through 2028. The partnership, valued at over $10 billion, aims to accelerate real‑time inference and improve response times for OpenAI’s customers. Cerebras’ low‑latency hardware will complement OpenAI’s existing compute portfolio, providing faster, more natural interactions for AI applications. Both companies highlighted the strategic fit, noting that the deal strengthens OpenAI’s infrastructure while showcasing Cerebras’ advanced chip technology. Leer más

OpenAI, led by Sam Altman, is partnering with designer Jony Ive to develop a palm‑sized, screen‑less AI gadget that can sit on a desk or be carried. The device will interact through microphones, speakers and cameras, aiming to offer a conversational experience that feels like a helpful friend rather than a generic digital assistant. Engineers are wrestling with challenges around the device’s voice, personality, engagement timing, privacy and the massive compute power needed to run advanced language models. The first unit is slated for release in late 2026 or early 2027. Leer más

Nvidia and OpenAI have struck a landmark partnership that could see Nvidia invest up to $100 billion in OpenAI while providing the massive compute infrastructure needed for the next generation of AI models. The agreement calls for OpenAI to deploy at least 10 gigawatts of Nvidia systems across new data centers, beginning with a one‑gigawatt rollout on Nvidia’s Vera Rubin platform in the second half of 2026. The deal positions Nvidia as both a supplier and a stakeholder, anchoring it at the heart of the AI boom and giving OpenAI a long‑term pipeline of cutting‑edge hardware to stay competitive with rivals. Leer más