Good morning and welcome back to another edition of The AI Rundown by Lightscape Partners.

• AMD’s purchase of software-optimization startup Brium is a direct strike at Nvidia’s moat: the deal lets AI models tuned for Nvidia GPUs run natively on AMD Instinct hardware, easing vendor lock-in for cloud and enterprise buyers and sharpening competition in the data-center AI market.

• A new UN report warns that Big Tech’s AI push is coming with a hidden cost: carbon emissions from Amazon, Microsoft, Alphabet, and Meta have jumped 150 % in three years. The finding spotlights the tension between relentless AI expansion and net-zero pledges, and it may push investors and regulators to scrutinize data-center buildouts far more closely.

• OpenAI revealed it has dismantled a network of at least ten covert influence operations—four allegedly linked to Chinese state actors—using ChatGPT for political messaging. The disclosure underscores both the power and the vulnerability of generative AI platforms, and it sets a precedent for proactive policing of nation-state misuse.

Stay tuned as we explore these stories and their implications for the future of AI, technology, and innovation.

If you haven’t yet, please support the newsletter by subscribing!

Hardware + Software

AMD acquires AI software startup Brium, aiming to challenge Nvidia’s data-center dominance. Link.

• AMD purchased AI software optimization firm Brium, whose specialized tools enable AI models originally optimized for Nvidia hardware to run efficiently on AMD’s Instinct GPUs and other processors, fostering hardware compatibility.

• The acquisition supports AMD’s goal of developing an open AI software ecosystem, directly targeting Nvidia’s market dominance in AI workloads by reducing software dependency on Nvidia-exclusive optimizations.

• Brium’s technology will significantly simplify developers’ ability to deploy AI applications across varied hardware environments, enhancing AMD’s appeal to enterprise customers building flexible, hardware-agnostic AI infrastructures.

• This strategic move highlights AMD’s determined push into the lucrative data-center AI market, emphasizing its intent to break Nvidia’s stranglehold by promoting widespread adoption of AMD-powered AI solutions.

Amazon developing humanoid delivery robots powered by advanced AI navigation systems. Link.

• Amazon is building advanced AI-driven software for humanoid delivery robots and testing them in an indoor "humanoid park" obstacle course at its San Francisco offices, initially relying on third-party robot hardware.

• These humanoid bots, equipped with Amazon’s sophisticated navigation AI, aim to address labor shortages and reduce delivery costs, complementing ongoing efforts to automate logistics operations across warehouses and final-mile delivery.

• The development signifies Amazon's expanding commitment to robotics and automation technologies, highlighting its ambition to streamline supply chains and significantly reshape the landscape of urban logistics.

• Successfully deploying these robots could revolutionize last-mile logistics, potentially transforming the economics of package delivery by drastically cutting reliance on human couriers and reducing operational expenses.

Figma launches AI-friendly design servers to accelerate coding workflows and streamline development. Link.

• Collaborative platform Figma unveiled a beta Dev Mode “Model Context Protocol” (MCP) server that enables AI coding tools, such as GitHub Copilot, to directly access and interpret design data, significantly speeding code generation.

• MCP servers expose design components, styles, and layout information through APIs, allowing generative AI to translate design mockups directly into production-ready frontend code, drastically improving workflow efficiency and accuracy.

• By automating the design-to-code transition, Figma aims to reduce manual developer workload and errors, enhancing developer productivity and promoting tighter collaboration between designers and engineers.

• This innovation underscores Figma’s proactive integration with AI-driven developer tools, positioning it to capture greater market share in developer-focused collaboration and AI-enabled software production.

Models

Google upgrades Gemini 2.5 Pro model, significantly enhancing coding capabilities. Link.

• Google announced major improvements to its Gemini 2.5 Pro AI model, specifically optimized to deliver superior performance in generating and debugging complex code, outperforming earlier versions on standard programming benchmarks.

• The updated model, accessible through Google's AI Studio and Vertex AI platforms, positions Gemini as a stronger competitor to OpenAI’s Codex, aiming to attract developers seeking robust and efficient AI coding assistants.

• This update is part of Google's strategic efforts to strengthen its foothold in enterprise-grade software development, leveraging Gemini’s multimodal capabilities to enhance AI-assisted software engineering tasks.

• Enhanced coding proficiency in Gemini 2.5 Pro demonstrates Google's continuing investment in AI to streamline development workflows, potentially redefining productivity standards for software development teams.

Consumer + Enterprise AI Applications

Anthropic launches Claude Gov, specialized AI assistant tailored for U.S. defense and intelligence agencies. Link.

• Anthropic introduced Claude Gov, a tailored version of its Claude AI assistant designed specifically for U.S. national security, intelligence analysis, and defense-related tasks, with fewer safety guardrails to handle sensitive classified content effectively.

• Optimized for threat assessment and military document analysis, Claude Gov provides government agencies with specialized AI capabilities, highlighting Anthropic's strategic bid for lucrative government contracts amid increased federal AI investment.

• This launch directly competes with OpenAI’s similar “ChatGPT Gov,” demonstrating Anthropic’s strategic effort to carve out market share in the growing sector of AI tools for defense and intelligence purposes.

• The tailored model underscores a significant trend toward customized enterprise-grade AI solutions, acknowledging the differing AI behavior requirements of government entities compared to private-sector customers.

Samsung introduces controversial AI-driven shopping assistant "Glance AI" on Galaxy lock screens. Link.

• Samsung launched "Glance AI," a lock-screen shopping app powered by generative AI, initially rolled out to Galaxy smartphones (S22 through S25), providing users with personalized product recommendations directly on their lock screens.

• Developed in partnership with mobile ad company InMobi, the feature, initially ad-free in the U.S., has raised user concerns about bloatware, privacy implications, and intrusion into a traditionally marketing-free space.

• While Samsung emphasizes innovative and frictionless shopping experiences, the deployment of Glance AI in the U.S. market tests consumer tolerance for invasive AI-driven marketing techniques embedded directly into everyday device usage.

• This initiative highlights an increasing trend toward integrating AI-driven monetization features directly into mobile operating systems, potentially redefining the relationship between users, brands, and platform providers.

Content Creation

DeviantArt founder introduces high-end $22,000 digital art display specifically designed for generative AI art. Link.

• Angelo Sotira, co-founder of DeviantArt, unveiled "Layer," a luxury digital frame priced at $22,000, meticulously engineered with premium hardware including 4K microLED panels to optimally display evolving generative art pieces.

• The premium price targets affluent art collectors, emphasizing a niche yet growing market segment interested in dynamic, generative AI art presented in high-quality, museum-grade digital frames.

• By providing cutting-edge display technology designed explicitly for generative AI visuals, Layer underscores growing recognition of AI-created art as serious collectible assets comparable in value to traditional physical artworks.

• This launch signals a strategic move within the art industry toward embracing advanced digital mediums, potentially accelerating mainstream acceptance of generative AI creations in elite artistic circles.

Data Centers + Energy

UN report finds tech giants' AI-driven carbon emissions have surged by 150% amid massive data center expansions. Link.

• The International Telecommunication Union (ITU) reported a 150% spike in carbon emissions over three years from Amazon, Microsoft, Alphabet, and Meta, driven primarily by energy-intensive AI data center operations.

• Amazon led with the highest increase (182%) in operational emissions, followed by Microsoft (155%), Meta (145%), and Alphabet (138%), reflecting extensive AI training and inference workloads significantly boosting electricity consumption.

• These soaring emissions challenge the tech giants' stated climate commitments, highlighting a growing industry dilemma: balancing rapid AI expansion against sustainable and environmentally responsible business practices.

• The UN calls for greater transparency and accelerated adoption of energy-efficient technologies as AI adoption expands rapidly, emphasizing urgent need for corporate action to mitigate environmental impact.

Startup Funding & Valuations

Elon Musk’s xAI aims for $300M raise at a staggering $113B valuation in secondary share sale. Link.

• Musk's AI startup xAI is launching a $300 million tender offer allowing employees to sell shares, valuing the company around $113 billion following its recent $33 billion acquisition of social platform X.

• The move enables xAI to bolster its cash reserves and support further product development after encountering delays in updates for its flagship AI model “Grok,” which has faced scrutiny over performance and safety issues.

• This massive valuation indicates substantial investor confidence in Musk's ability to position xAI competitively against industry giants like OpenAI, Microsoft, and Google amid intensifying AI rivalry.

• The tender offer highlights a strategic financial maneuver to sustain momentum and incentivize employee retention during a period of aggressive expansion and integration of Musk’s diverse tech ventures.

Space Forge raises $30M Series A to manufacture high-performance semiconductor wafers in space. Link.

• UK-based Space Forge secured approximately $30 million led by the NATO Innovation Fund and Northrop Grumman to produce advanced semiconductor wafers in microgravity conditions aboard orbiting spacecraft.

• The microgravity environment significantly reduces defects in semiconductor crystal structures, enhancing chip efficiency and reducing power consumption, with potential uses in critical sectors like 5G networks and quantum computing.

• Demonstrating feasibility, Space Forge already conducted tests with telecom giant BT, confirming the improved performance of chips manufactured in space compared to traditional Earth-based production methods.

• This investment underscores growing interest from both commercial and defense sectors in exploiting space conditions for advanced manufacturing, reflecting an emerging market opportunity for microgravity-enabled industrial processes.

Taiwan’s Wistron seeks nearly $1B share sale to ramp up AI hardware production. Link.

• Taiwan-based Wistron, a key electronics supplier to Nvidia, plans to raise up to $923 million via global depositary share sales, aiming to finance the procurement of raw materials essential for increased AI hardware manufacturing.

• The offering, expected at a 4–6% discount to current market pricing, will bolster production capabilities, especially with Wistron's forthcoming U.S.-based facilities intended to produce high-performance computing and AI server hardware.

• Growing demand for advanced AI servers and data center equipment is driving Wistron’s aggressive expansion, underpinned by surging global investments into generative AI and related computing infrastructure.

• This capital raise exemplifies broader industry trends of significant investments by tech manufacturers positioning themselves to capture the explosive growth of the global AI hardware market.

OpenAI

ChatGPT adds file-search capabilities as OpenAI’s enterprise user base exceeds 3 million. Link.

• OpenAI announced that ChatGPT now supports querying internal enterprise files from major cloud platforms such as Google Drive, Dropbox, Box, SharePoint, and OneDrive, significantly boosting its utility as a business productivity tool.

• OpenAI’s enterprise customer base has rapidly grown to over 3 million paying business users, up from 2 million in February 2025, underscoring the rapid adoption of ChatGPT in corporate workflows.

• The update includes a new "record mode" feature that transcribes and summarizes meetings, reinforcing ChatGPT’s role as an essential business tool designed to streamline internal communication and document access.

• These enhancements reflect OpenAI’s aggressive push into enterprise markets, positioning ChatGPT as a comprehensive AI-powered business assistant amid intensifying competition from Microsoft, Google, Amazon, and Anthropic.

OpenAI dismantles covert propaganda networks exploiting ChatGPT for influence operations. Link.

• OpenAI revealed it disrupted at least ten covert propaganda networks misusing its AI tools, including four believed to be operated by Chinese state actors employing ChatGPT-generated content to disseminate pro-Beijing narratives.

• The detected operations involved using ChatGPT to craft social media posts, comments, and even internal performance reports, highlighting how generative AI tools can be weaponized for sophisticated geopolitical influence campaigns.

• This marks OpenAI’s first public acknowledgment of nation-state abuse of its technology, prompting strengthened safeguards to curb similar exploitation in the future, particularly concerning politically sensitive content.

• The disclosure underscores escalating challenges AI firms face in safeguarding their technologies from malicious use, highlighting the urgent need for robust security measures and proactive monitoring to protect against digital manipulation.

Thank you for reading the AI Rundown by Lightscape Partners. Please send any questions, comments, or suggestions to [email protected].