Starcloud, an NVIDIA-backed startup, recently achieved a significant milestone by successfully training its first AI model in space. The Starcloud-1 satellite, launched on November 2, 2025, utilized an NVIDIA H100 GPU to run Google's Gemma large language model and train Andrej Karpathy's nano-GPT on Shakespeare's works. CEO Philip Johnston highlighted the advantages of orbital data centers, such as constant solar power and passive cooling, which could lead to energy costs 10 times lower than Earth-based facilities. Starcloud plans to expand this initiative, aiming for a 5 GW orbital cluster equipped with NVIDIA H200 chips by October 2026. The demand for AI infrastructure continues to surge, with Coreweave CEO Michael Intriligator describing it as a "violent change." Coreweave, a key partner of Nvidia, is rapidly expanding its data center capacity to provide GPUs to AI companies. While Nvidia's GPU ecosystem remains a standard for modern AI computing, Google's latest TPU advancements are challenging this dominance. Google's TPUs show potential to outperform GPUs in certain large language model tasks, intensifying the competition in AI processing technology. Globally, discussions around AI governance and intellectual property are intensifying. An Indian government panel proposed that AI companies, including OpenAI and Google, pay content creators for using their work to train AI models, a stance differing from the US "fair use" approach. This proposal, released on December 9, 2025, suggests a central body for royalty collection, though tech industry group Nasscom, which includes Google and Microsoft, opposes it as a "tax on innovation." Meanwhile, the U.S. Department of Health and Human Services (HHS) released its 21-page AI strategy on December 4, 2025, focusing on responsible AI use across its divisions, with the FDA launching its own "agentic AI" platform on December 1, 2025. Security professionals also anticipate strict questions about AI governance, data sovereignty, and human oversight from major buyers starting in 2026. AI's influence is reshaping various sectors, from entertainment to finance. Actress and filmmaker Natasha Lyonne, speaking at Fortune's Brainstorm AI conference on December 10, 2025, expressed deep concern for Hollywood crews and unions, viewing AI as a "fear tactic" that threatens jobs. She advocates for broader participation in AI's development and uses an AI video model called Marey for a new film, blending AI with human filmmaking. In finance, Goldman Sachs CFO Denis Coleman detailed the company's OneGS 3.0 initiative, a multiyear plan to integrate AI across all operations to boost productivity and drive growth, expecting a net increase in staff by year-end 2025. New platforms and protocols are emerging to enhance AI capabilities and interoperability. Stagwell launched NewVoices.ai on December 10, 2025, an adaptive AI platform for enterprise sales, support, and retention that offers personalized, one-to-one interactions. Furthermore, major AI companies are adopting the Model Context Protocol (MCP), developed by Anthropic in mid-2024 and donated to the Linux Foundation. This protocol allows AI agents to access internet information and tools in a standardized way. OpenAI, Google, and Microsoft are joining the Agentic AI Foundation to advance open-source agentic AI, enabling systems like Anthropic's Claude to seamlessly connect with services such as Slack.
Key Takeaways
- Starcloud, backed by Nvidia, successfully trained its first AI model in space on November 2, 2025, using an Nvidia H100 GPU and Google's Gemma LLM.
- Starcloud aims for a 5 GW orbital cluster with Nvidia H200 chips by October 2026, projecting 10 times lower energy costs than Earth-based data centers.
- Coreweave CEO Michael Intriligator defended AI deals with Nvidia, noting "violent change" in demand for AI infrastructure as Coreweave expands GPU capacity.
- Google's latest TPUs are challenging Nvidia's dominance in AI hardware, potentially outperforming GPUs in some large language model tasks.
- An Indian government panel proposed that AI companies like OpenAI and Google pay content creators for training data, a plan opposed by Nasscom (including Google and Microsoft).
- The U.S. Department of Health and Human Services (HHS) released a 21-page AI strategy on December 4, 2025, focusing on responsible AI use across its divisions.
- Actress Natasha Lyonne, speaking on December 10, 2025, expressed concern over AI's threat to Hollywood jobs and advocates for broader voice in AI development, while using an AI video model for her own film.
- Goldman Sachs CFO Denis Coleman detailed the OneGS 3.0 initiative to integrate AI across all operations, aiming to boost productivity and expecting a net increase in staff by year-end 2025.
- Stagwell launched NewVoices.ai on December 10, 2025, an adaptive AI platform for enterprise sales, support, and retention, offering personalized interactions.
- Anthropic developed the Model Context Protocol (MCP) in mid-2024, donating it to the Linux Foundation, with OpenAI, Google, and Microsoft joining the Agentic AI Foundation to advance open-source agentic AI.
Starcloud trains AI model in space with Nvidia H100
Starcloud, backed by Nvidia, successfully trained its first AI model in space. The Starcloud-1 satellite launched on November 2, 2025, with an Nvidia H100 GPU. It ran Google's Gemma large language model and trained NanoGPT on Shakespeare's works. CEO Philip Johnston believes orbital data centers will have 10 times lower energy costs than Earth-based ones. This project shows space can host powerful AI computing, aiming to reduce terrestrial energy strain.
Starcloud uses Nvidia hardware to train AI in space
Starcloud successfully trained its first AI model in space using Nvidia hardware. The Starcloud-1 satellite, launched in early November 2025, carries an Nvidia H100 GPU. CEO Philip Johnston stated it ran Google's Gemma and trained NanoGPT on Shakespeare's works. Orbital data centers could cut energy costs by 10 times compared to Earth. Starcloud plans a 5 GW orbital cluster with Nvidia H200 chips by October 2026, powered by solar energy.
Starcloud trains first AI model in space with Nvidia
Starcloud, an NVIDIA-backed startup, successfully trained the first large language model in space. Its Starcloud-1 satellite used an NVIDIA H100 GPU to train Andrej Karpathy's nano-GPT model on Shakespeare's works. CEO Philip Johnston said space offers advantages like constant solar power and passive cooling. This could make energy costs 10 times cheaper and reduce carbon emissions. Starcloud aims for gigawatt-scale orbital data centers by the early 2030s.
Natasha Lyonne discusses AI future in Hollywood
Actress and filmmaker Natasha Lyonne spoke at Fortune's Brainstorm AI conference on December 10, 2025. She believes AI, like Tilly Norwood, is a "fear tactic" and not the true future of Hollywood. Lyonne expressed deep concern for crews, cast, and unions like IATSE and Teamsters, whose jobs are threatened by AI. She highlighted the tension between executives and creative workers regarding AI's role. Lyonne also noted that even if artists avoid AI, their marketing departments likely use it.
Natasha Lyonne shapes AI future in Hollywood
Actress Natasha Lyonne, co-founder of Asteria Film Co., spoke at Fortune's Brainstorm AI conference on December 10, 2025. She questioned why AI focuses on replacing screenwriters instead of solving problems like ocean plastic pollution. Lyonne, named one of TIME's 100 Most Influential People in AI 2025, advocates for more people to have a voice in AI's development. Her company Uncanny Valley uses an AI video model called Marey, built on licensed data, for a new film that blends AI and human filmmaking. She emphasizes mastering skills to build new opportunities in the AI era.
Coreweave CEO defends AI deals with Nvidia
Coreweave CEO Michael Intriligator defended the company's AI deals with Nvidia as "working together" on Yahoo Finance Live. He explained that Coreweave provides GPUs to AI companies, and Nvidia is a key partner. Intriligator described the demand for AI infrastructure as a "violent change" due to unprecedented growth. Coreweave is rapidly expanding its data center capacity to meet this demand. The company has secured significant funding and long-term contracts, focusing on high-performance computing at competitive prices.
India proposes AI companies pay content creators
An Indian government panel proposed that AI companies like OpenAI and Google pay content creators for using their work to train AI models. This plan differs from the US, where AI firms claim "fair use" for public data. India suggests a central body collect royalties, calling the EU's opt-out model ineffective. Nasscom, a tech industry group including Google and Microsoft, opposes this, calling it a "tax on innovation." The public has 30 days to challenge the proposal, which was released on December 9, 2025.
Google TPUs challenge Nvidia in AI hardware
Google's latest TPU advancements are challenging Nvidia's dominance in the AI hardware market. While TPUs may outperform GPUs in some large language model tasks, industry experts note that Nvidia's complete GPU ecosystem remains the standard for modern AI computing. This competition highlights the ongoing race in AI processing technology.
Goldman Sachs CFO discusses AI driven growth plan
Goldman Sachs CFO Denis Coleman discussed the company's OneGS 3.0 initiative, a multiyear plan to integrate AI across all operations. This overhaul aims to reduce complexity, boost productivity, and drive growth. The company is asking employees to rethink processes and is investing in AI to accelerate change. Despite headcount reductions in early 2025, Goldman Sachs expects a net increase in staff by year-end. Coleman described the US economic outlook as resilient and conducive to business.
HHS unveils AI strategy for health innovation
The U.S. Department of Health and Human Services (HHS) released a 21-page AI strategy on December 4, 2025. Led by Acting Chief AI Officer Clark Minor, the plan aims to use AI responsibly across the department. It focuses on five pillars: governance, infrastructure, workforce development, health research, and care delivery. This "OneHHS" approach includes sharing AI models across divisions like the FDA, CDC, CMS, and NIH. Just before this, on December 1, 2025, the FDA launched its own agency-wide "agentic AI" platform for employees, featuring human oversight.
Security experts face tough AI questions in 2026
Security professionals must prepare for strict questions about AI governance starting in 2026. Major buyers, especially in government and critical infrastructure, now deeply understand AI risks. They will demand details on generative AI controls, including data sovereignty, human oversight, and compliance with data protection and intellectual property laws. Many organizations currently use AI informally, lacking structured governance for client data, output checks, and incident tracking. This lack of clear policies creates risks in areas like data protection, information security, and professional indemnity.
Stagwell launches NewVoices.ai for enterprise sales
Stagwell (STGW) launched NewVoices.ai, an adaptive AI platform for enterprise sales, support, and retention. Announced on December 10, 2025, this platform offers personalized, one-to-one interactions that continuously learn from user preferences. Mark Penn, Stagwell's CEO, stated it opens a huge new market for agent interaction. NewVoices.ai, founded by Eran Nizri, uses Stagwell's proprietary data to deliver human-level tailored experiences. It provides solutions for sales, customer support, payments, and feedback, aiming to disrupt the revenue ecosystem.
AI companies adopt MCP for new internet
Major AI companies are adopting Model Context Protocol (MCP) to build the next generation of apps and services. MCP allows AI agents to access information and tools across the internet in a standardized way. Anthropic, which developed MCP in mid-2024, is donating it to the Linux Foundation. Companies like OpenAI, Google, and Microsoft are also joining the Agentic AI Foundation to advance open-source agentic AI. This protocol, likened to new APIs, enables AI systems like Claude to seamlessly connect with services such as Slack, creating a "ping-pong of intelligence" for users.
Sources
- ‘Greetings, earthlings’: Nvidia-backed Starcloud trains first AI model in space as orbital data center race heats up
- Starcloud trains first AI model in space using Nvidia hardware
- Space Datacenter Startup Starcloud Successfully Trains First LLM In Space
- Natasha Lyonne: Tilly Norwood isn’t the real future of AI in Hollywood
- Actress Natasha Lyonne dropped out of NYU and watched movies at the Film Forum instead. Now, she’s helping to shape the future of AI.
- Coreweave CEO defends AI circular deals as ‘working together’
- Indian AI royalty proposal targets data practices of OpenAI, Google
- Google TPU gains test Nvidia's lead, but the GPU ecosystem still rules
- Goldman Sachs CFO on the company's AI reboot, talent, and growth
- HHS Releases Strategy Positioning Artificial Intelligence the Core of Health Innovation | Insights
- Security pros should prepare for tough questions on AI in 2026
- Stagwell (STGW) Launches NewVoices.ai - An Enterprise Sales, Support and Retention Platform Powered by Adaptive AI
- AI companies want a new internet — and they think they’ve found the key
Comments
Please log in to post a comment.