Silism Week 9 (2026): Samsung integrates Perplexity, Nvidia eyes $30B OpenAI investment, Anthropic clashes with the Pentagon, Meta signs 6GW AI deal with AMD, and enterprise AI infrastructure accelerates.

AI power is shifting.

Not just between models — but between:

• Governments and labs
• Chipmakers and hyperscalers
• Infrastructure and enterprise
• Safety and surveillance

Last week:

• Nvidia eyes $30B into OpenAI.
• Anthropic rejects Pentagon demands.
• Samsung embeds Perplexity at the OS level.
• Meta signs 6GW inference with AMD.
• Goldman says AI added “basically zero” to GDP.

Meanwhile, enterprise automation accelerates quietly.

Week 9 of Silism breaks down the structural moves behind the headlines.

If you care about where AI power is consolidating — not just which model is trending — subscribe.

Comment what surprised you most last week.

Need-to-Know

• Samsung is embedding Perplexity into Galaxy AI on the Galaxy S26, building a multi-agent system with deep OS-level app access.
• YouTube is piloting a conversational AI chatbot across TVs, consoles, and streaming devices to let viewers ask questions mid-watch.
• OpenAI is reportedly testing a $100/month ChatGPT Pro Lite plan to bridge the gap between Plus and Pro tiers.
• GenAI.mil, the Pentagon’s generative AI platform, partnered with OpenAI to scale AI-powered military data analysis and training.
• The Guardian reported Nvidia may invest $30B in OpenAI’s next round after a previous $100B arrangement dissolved.
• Amazon pledged $12B for AI data centers in Louisiana and will fund related energy and water infrastructure.
• xAI reached a Pentagon agreement to deploy Grok in classified systems under an “all lawful use” framework.
• IBM shares dropped ~13% after Anthropic claimed Claude Code can modernize COBOL, threatening legacy mainframe demand.
• Goldman Sachs said AI investment added “basically zero” to U.S. GDP in 2025, as imports offset productivity gains.
• Guide Labs open-sourced Steerling-8B, an interpretable LLM with token-level traceability via a built-in concept layer.
• Anthropic upgraded Claude Cowork to integrate with Google Workspace and automate Excel and PowerPoint workflows.
• AMD signed a 6GW AI inference agreement with Meta, including equity warrants, echoing its OpenAI deal structure.
• Inception Labs introduced Mercury 2, a diffusion-based reasoning model positioned as a faster, cheaper alternative to Transformers.
• Nimble launched Agentic Search and raised $47M Series B to enable enterprise AI to fetch and validate live web data.
• The Pentagon pressured Anthropic for broader model access under DPA threat, but the company refused.
• Anthropic rolled out Remote Control for Claude Code, enabling users to start tasks locally and manage them via mobile.
• DeepSeek reportedly withheld V4 from Nvidia and AMD, favored Huawei, and faced allegations around Blackwell chips.
• Trump stated tech firms would pledge power commitments for AI data centers, though specifics remain unclear.
• Adobe launched Quick Cut in Firefly, converting raw or AI-generated footage into rough edits from text prompts.
• Google relaunched Flow, unifying text-to-image and text-to-video tools into one creative workflow.
• Anthropic acquired Vercept, a cloud computer-use agent startup, and integrated its team.
• Anthropic rejected Pentagon demands for unrestricted access, opposing surveillance and autonomous weapons use.
• Microsoft previewed Copilot Tasks, a cloud AI handling emails, subscriptions, and scheduling.
• Anthropic added auto-memory to Claude Code to persist project context and preferences.
• Burger King launched Patty, an AI headset assistant for meal prep and service interaction monitoring.
• Mistral AI signed a multiyear enterprise AI partnership with Accenture.

---

Weekly AI Launches

• Rork Max released a web-based Swift app builder for Apple platforms with one-click installs.
• git-lrc launched a pre-commit AI code reviewer focused on bugs and security risks.
• WANotifier relaunched as a zero-markup WhatsApp CRM built on Meta’s official API.
• Prism Videos introduced a centralized cloud AI video timeline editor.
• Straion launched a centralized rule manager feeding context to enterprise AI coding agents.
• Tidy launched a cloud personal agent that navigates websites and reports via iMessage.
• Wordy introduced a movie-based vocabulary learning app with quizzes.
• PulseKit launched an Apple widget dashboard for live business metrics.
• Modelence launched an open-source-backed rapid app builder.
• Anima launched a Figma-to-code AI for responsive frontend UI.
• Foxchat introduced a lightweight site chat widget routing to Slack.
• Orchids launched an app builder using existing AI subscriptions.
• Arzule released an AI platform replacing spreadsheets for B2B SaaS partnership optimization.
• floors.js launched a script converting websites into multiplayer 3D spaces.
• Ask Fellow launched an AI meeting assistant with export and follow-up automation.
• DemoMe released an on-device product demo video builder.
• ChatPal introduced a conversation-first AI language learning app.
• Koidex launched a security scanner for code packages and AI models.
• Rover launched a website AI agent capable of executing user actions.
• Tessl released an AI agent evaluation platform to reduce bugs and hallucinations.

If you care about where AI power is consolidating — not just which model is trending — subscribe.

Comment what surprised you most last week.