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

• Defense startup Anduril is raising roughly $4 billion in new funding that could push its valuation toward $60 billion. The round reflects surging investor interest in defense AI as autonomous systems move from prototypes to battlefield deployments, particularly with Anduril’s drones already operating in Ukraine.

• OpenAI’s Pentagon partnership triggered consumer backlash even as ChatGPT usage keeps climbing. Weekly users topped 800 million, but U.S. app deletions reportedly jumped 295 percent after news of the defense agreement spread, highlighting how military work can create brand risk for consumer AI products.

• Netflix acquired Ben Affleck’s AI filmmaking startup InterPositive to improve postproduction and editing workflows. The technology focuses on processing footage for tasks like color correction and effects, signaling that studios currently prefer operational AI that speeds production rather than tools that replace creative roles.

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

Broadcom says AI chip sales could top $100 billion next year as hyperscalers race to secure custom silicon capacity. Link.

• Reuters reports Big Tech plans to spend more than $600 billion on AI infrastructure in 2026, lifting chip demand sharply.

• Broadcom says it sees about 10 gigawatts of 2027 AI chip demand from customers including Meta and Anthropic.

• The company also says it has locked in advanced wafer and memory supply through 2028 to meet forecasted orders.

• The outlook strengthens Broadcom's case as a serious Nvidia challenger in custom AI accelerators and networking infrastructure.

Content Creation

Netflix bought Ben Affleck's AI filmmaking startup InterPositive to improve postproduction and editing workflows. Link.

• The Verge reports all 16 InterPositive engineers and researchers will join Netflix, while Affleck becomes a senior advisor.

• InterPositive focuses on processing production footage for editing, color correction, and effects rather than generating scripts.

• Affleck framed the technology as a way to remove tedious work so creatives can focus on performance and storytelling.

• The deal shows studios still favor operational AI that improves execution over tools that replace core human creativity.

Consumer AI Applications

MyFitnessPal acquired the teen-built Cal AI app after its photo-based calorie counter exploded in downloads and revenue. Link.

• TechCrunch says Cal AI uses computer vision to estimate meal calories and reached about 15 million downloads.

• The startup reportedly grew to around $30 million in annual revenue before the acquisition by MyFitnessPal.

• MyFitnessPal plans to keep Cal AI as a standalone product while integrating it with its nutrition database.

• The deal highlights strong consumer demand for lightweight AI tools that turn image recognition into daily utility.

Apple introduced optional AI labels in Apple Music so providers can disclose when songs or visuals use generative tools. Link.

• The Verge says Apple will assume no AI was used if labels are not applied by music providers.

• The change follows growing industry pressure for transparency as synthetic and hybrid content spreads across music platforms.

• The system applies to both songs and visualizer-style assets connected to music releases inside Apple Music.

• Labeling gives Apple a way to address trust concerns without banning AI-assisted creative work outright.

Data Centers + Energy

Oracle and OpenAI abandoned a major Texas data center expansion tied to the Stargate infrastructure plan. Link.

• Reuters says the cancelled build would have added roughly 600 megawatts near the main Abilene supercomputing site.

• The wider Stargate initiative was pitched as a $500 billion and 10 gigawatt AI infrastructure program.

• Oracle and OpenAI will shift future capacity to other locations even as the existing Texas campus keeps ramping.

• The reversal highlights how financing and power strategy can still derail even the biggest AI infrastructure ambitions.

Startup Funding & Valuations

Anduril is raising about $4 billion in fresh capital, nearly doubling its valuation as defense AI demand surges. Link.

• Reuters reports Thrive Capital and Andreessen Horowitz are leading the round, pushing Anduril's valuation toward $60 billion.

• The company has become one of Silicon Valley's highest profile defense bets as autonomous systems reshape military procurement.

• Anduril's AI-enabled drones are already deployed in Ukraine, reinforcing investor confidence in battlefield-tested products.

• The round shows private capital still flooding defense tech even as other growth sectors face harsher funding scrutiny.

Lio raised $30 million to automate procurement with AI agents that can negotiate, compare, and complete purchasing tasks. Link.

• TechCrunch says Andreessen Horowitz led the round, bringing Lio's total funding to roughly $33 million.

• The company says its agents connect to enterprise systems and can handle documents, supplier data, and transaction workflows.

• One pilot customer reportedly automated most outsourced procurement activity within six months using Lio's platform.

• The pitch reflects growing demand for AI systems that do work directly, not just assist employees inside dashboards.

SolveAI landed $50 million to build enterprise coding tools tailored to each company's internal software environment. Link.

• Fortune says the startup was founded in July and quickly raised a seed round before closing a larger Series A.

• SolveAI's product aims to generate code that matches internal engineering context rather than generic public patterns.

• The company is positioning itself against a crowded field of AI coding assistants and enterprise developer platforms.

• Backer interest shows investors still see room for new winners in software engineering automation and code generation.

Ease Health emerged from stealth with $41 million to build an AI-native operating system for behavioral health providers. Link.

• The company says its platform unifies admissions, clinical records, and billing into a single operational software stack.

• Andreessen Horowitz led the Series A, signaling continued investor appetite for healthcare infrastructure with embedded AI automation.

• Behavioral health providers face administrative burdens that create an opening for software promising workflow efficiency gains.

• The funding will support product expansion in a sector where staffing pressure and fragmented systems remain major constraints.

Regulation + Legal

The Pentagon labeled Anthropic a supply-chain risk after the company refused broad military use of Claude. Link.

• The Verge reports the designation bars defense contractors from using Anthropic systems and escalates the Pentagon dispute.

• Anthropic says it will challenge the move in court and argues the government is punishing the company for safety limits.

• The label is unusual because it is typically associated with foreign adversary-linked vendors, not a domestic AI startup.

• The clash raises major questions about procurement pressure and how much government leverage labs can resist.

A federal judge dismissed xAI's suit accusing OpenAI of poaching employees to steal trade secrets. Link.

• Business Insider says the court found xAI had not presented specific misconduct by OpenAI itself in the complaint.

• The ruling gives xAI time to amend its filing, but represents an early legal win for Sam Altman's company.

• The case is one front in Elon Musk's wider fight with OpenAI over talent, competition, and corporate direction.

• It shows how AI rivalry is moving beyond products and funding into lawsuits over people and information flow.

Safety + Ethics

Anthropic's CEO reportedly condemned OpenAI's public framing of its military deal, deepening the industry split over defense AI. Link.

• TechCrunch says a leaked memo shows Dario Amodei calling OpenAI's safety messaging around the contract straight up lies.

• The dispute centers on whether frontier labs should accept Pentagon work without stricter limits on weapons and surveillance use.

• Anthropic had refused broader military terms, while OpenAI moved ahead with a classified deployment agreement.

• The memo underscores how safety positioning is becoming both a governance issue and a competitive branding battle.

OpenAI

ChatGPT usage is still growing rapidly, but OpenAI's Pentagon deal triggered a sharp spike in app deletions and backlash. Link.

• TechCrunch says weekly active users topped 800 million even after a post-holiday slowdown earlier in the year.

• US app uninstalls jumped 295 percent after news of OpenAI's Department of Defense agreement spread across social platforms.

• Claude downloads rose as some users shifted to Anthropic, while ChatGPT accumulated a wave of one-star reviews.

• The episode shows how military partnerships can create real consumer brand risk even for dominant AI products.

OpenAI launched GPT-5.4 with Pro and Thinking variants, pushing harder on professional workflows and reasoning depth. Link.

• TechCrunch reports the model supports context windows up to one million tokens, far beyond earlier OpenAI releases.

• OpenAI says GPT-5.4 reduces factual errors versus GPT-5.2 and posts stronger benchmark performance on difficult tasks.

• The company also introduced Tool Search so the API can dynamically retrieve tool definitions during execution.

• The release suggests frontier model competition is shifting toward practical work systems instead of simple benchmark bragging rights.

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