AI keeps expanding in scale, with gigawatt campuses, nation-level investments, and record funding rounds. Yet behind every breakthrough is the same engine that has always driven technology forward: people who choose to build.
This week makes that clear. Countries are competing not only for chips but for the teams that know how to use them. Saudi Arabia’s HUMAIN forms a new venture with AMD and Cisco. HUMAIN also deepens its work with NVIDIA to stand up hundreds of thousands of GPUs. Foxconn teams up with OpenAI to rethink how advanced hardware is designed and manufactured. In the enterprise, Google’s Gemini 3 brings multimodality into daily workflows, Target embeds ChatGPT throughout its business, and OpenAI turns ChatGPT into a shared workspace where groups can collaborate with a model as selective participants. Even Grok 5’s delay reflects a human choice to trade speed for capability.
Capital continues to follow talent. NVIDIA and Microsoft anchor a massive expansion for Anthropic. Databricks climbs toward a new valuation peak. Luma AI raises $900 million. Adobe acquires Semrush. Jeff Bezos steps in to help shape a new AI company focused on engineering and manufacturing. Investors are aligning behind teams with the ambition and expertise to push the frontier forward.
Research and policy highlight the same theme. GPT-5 demonstrates potential in scientific acceleration while still relying on human judgment. AI2 releases the Olmo 3 family. DeepMind builds a new research hub in Singapore. Leading Meta researchers join fast-rising labs. Policymakers on both sides of the Atlantic adjust rules to reflect the expanding role of AI in work, security, and innovation.
On the global stage, the race is increasingly about who can attract and empower the right people. The United States and Saudi Arabia formalize a broad AI partnership. Nokia boosts its presence in the United States with a significant investment. Taiwan moves to protect its systems from foreign influence. Governments are not just competing for infrastructure. They are competing for the human capability to shape it and steer it.
AI keeps accelerating. What stands out this week is how much of the story leads back to the builders, operators, and decision makers who turn technology into progress.
Here’s your Saturday guide to the signals shaping the future of AI:
Infrastructure
AMD, Cisco, and HUMAIN created a joint venture to build 1 GW of AI infrastructure in Saudi Arabia, starting with a 100 MW site, and early tenants already signed, alongside HUMAIN’s separate plan with NVIDIA to deploy up to 600,000 GPUs across Saudi Arabia and the U.S. Click here
HUMAIN and NVIDIA expand their partnership to deploy up to 600,000 NVIDIA GPUs over three years, supporting AI infrastructure across Saudi Arabia and the U.S., training new models like HUMAIN Chat, and boosting compute for frontier AI systems. Click here
Foxconn is partnering with OpenAI to co-design AI hardware, including data-center racks and components, bolstering U.S.-aligned supply chains for advanced AI servers as export controls tighten on China. Click here
Enterprise
Google launches Gemini 3, its most advanced multimodal AI model yet, featuring PhD-level reasoning, native multimodality across text, image, audio, and video, and deep integration into Search and Workspace. Click here
OpenAI is rolling out group chats in ChatGPT, letting up to 20 people collaborate in the same conversation. At the same time, the model selectively joins in, supports mentions, reacts with emoji, and runs on GPT-5.1 Auto for responses. Click here
Tim Cook’s expected retirement has intensified focus on Apple’s AI future, as the company faces pressure to catch rivals in the AI race and is reportedly shifting resources toward AI-powered smart glasses; hardware chief John Ternus remains the leading successor. Click here
Elon Musk says xAI is delaying the launch of Grok 5 until next year, claiming the extra training time will make it the most capable AI model in the world and a major leap toward human-level intelligence. Click here
Target is adopting OpenAI’s ChatGPT across its business, adding ChatGPT to its shopping app for multi-item baskets and personalized recommendations while giving employees ChatGPT Enterprise for tasks like summarizing reports and drafting marketing content. Click here
Capital Flows
NVIDIA and Microsoft will invest up to $15 billion in Anthropic through a new cloud partnership, with Anthropic committing to $30 billion in Azure compute powered by NVIDIA’s next-gen systems and expanding Claude access across all major clouds. Click here
Databricks is in talks to raise capital at a valuation above $130 billion, a roughly 30% jump from the $100 billion valuation just two months ago, as it continues to ride the wave of AI and data analytics growth. Click here
Luma AI raised $900 million in a Series C round led by HUMAIN (backed by Saudi Arabia’s PIF), reinforcing the link between massive capital bets and sovereign compute infrastructure as the startup pursues multimodal AGI with global ambitions. Click here
Adobe will acquire Semrush for $1.9 billion in cash, adding the brand-visibility and GEO platform to its marketing and agentic-AI stack as demand grows for tools that keep brands discoverable across AI search, LLMs, and traditional channels. Click here
Jeff Bezos has become co-CEO and co-founder of Project Prometheus, a new AI startup that has raised $6.2 billion, positioning the venture to apply AI to engineering and manufacturing while quietly building a growing Bay Area presence. Click here
Research
OpenAI published early research showing how GPT-5 can accelerate scientific discovery, highlighting case studies where the model helped generate proofs, surface new mechanisms in biology, and speed up literature review—while noting limitations and the need for expert oversight. Click here
The Allen Institute for AI released its new Olmo 3 family of open models, including 7B and 32B-parameter base, Instruct and “Think” variants, with the 32B-parameter Olmo 3-Think becoming the first open-source reasoning model of its size and fully transparent training recipe. Click here
Google DeepMind has opened a new AI research lab in Singapore to serve the Asia-Pacific region, bringing together scientists, engineers, and policy experts to advance models like Gemini, collaborate with regional institutions, and focus on multilingual and culturally inclusive AI. Click here
Top Meta AI Researcher, Soumith Chintala, has joined Mira Murati’s startup Thinking Machines Lab, marking a major talent win for the fast-growing AI lab as it hires top researchers and prepares a new funding round reportedly targeting a $50 billion valuation. Click here
Policy
House Republicans are advancing a plan to override state AI laws through a national defense bill, aiming to establish a unified federal framework for AI regulation and pre-empt state-level rules as the federal government pushes for centralized policy. Click here
The European Commission introduced the “Digital Omnibus” proposal, aiming to streamline EU digital rules by delaying the implementation of high-risk AI system regulations and amending the GDPR to permit AI model training on personal data under “legitimate interest”. Click here
The bipartisan AI Infrastructure Coalition launched to push pro-innovation AI policies, backed by tech and energy companies along with figures like Kyrsten Sinema, aiming to promote federal build-out of data centers and infrastructure, reduce permitting hurdles, and counter fragmented state-level regulation. Click here
Global AI Strategy
The U.S. and Saudi Arabia signed a new AI agreement, giving Saudi Arabia access to U.S. tech systems and committing both countries to deeper cooperation on advanced chips, secure compute, and critical minerals as Saudi Arabia prepares to spend $50 billion on semiconductors. Click here
Nokia will invest $4 billion in the U.S., including $3.5 billion for AI R&D and $500 million for manufacturing, as it aligns its strategy with U.S. tech and supply-chain priorities—underscoring how AI is becoming a core pillar of national industrial policy and global geopolitical alignment. Click here
Taiwan has imposed a government-wide ban on Chinese-influenced AI systems, citing cybersecurity and disinformation risks linked to the Chinese Communist Party and China’s expanding digital footprint. Click here
📱Social Signals
The most important conversations in AI are unfolding across social media, where top voices are shaping the next wave of signals and strategy. Here are some of the top social signals and their takes from the past week.
Sundar Pichai (Click here) — “Introducing Gemini 3. It’s the best model in the world for multimodal understanding, and our most powerful agentic + vibe coding model yet. Gemini 3 can bring any idea to life, quickly grasping context and intent so you can get what you need with less prompting. Rolling out today across the Gemini app and AI Mode in Search, and available to developers in Google AI Studio and Vertex AI.” Pichai frames Gemini 3 as a shift from model to an agent, emphasizing natural contextual understanding and reduced prompt dependency. The post positions Gemini 3 as a leap toward conversational execution—where intent is inferred rather than engineered—marking a progression from AI tool to AI collaborator in both consumer and developer workflows.
Pascal Bornet (Click here) — “AI does not record meetings. It interprets them… Whoever controls the memory, controls the narrative.” Bornet argues that AI assistants in team check-ins shift from passive note-takers to active framers of meaning. He warns that perfect summaries can misrepresent intent by omitting nuance, silence, and emotional weight — influencing how teams remember decisions and shaping future behavior. His signal reframes AI not as a helper but as a “memory architect,” cautioning leaders to safeguard narrative authority and contextual judgment to avoid teams unconsciously adapting their thinking to match how AI listens rather than how humans work.
Elon Musk (Click here) — “You should notice rapid improvements in the quality of your timeline almost every week. We are also switching advertising recommendations to use the same Grok/AI system as organic posts, so quality of ads will get radically better.” Musk signals that X is moving toward a fully AI-optimized feed, where Grok powers both content discovery and monetization. He frames the shift as iterative (“two steps forward, one step back”) but positions AI-driven ad targeting as a breakthrough in relevance and user experience, blurring the line between organic and paid distribution.
Andrew Ng (Click here) — “AI is amazingly intelligent… and still incredibly dumb. I wouldn’t trust a frontier LLM to prioritize my calendar or screen resumés without extensive customization.” Ng responds to an 18-year-old anxious that AI will eliminate meaningful work before he enters the field, arguing that hype is distorting reality. He emphasizes that AI remains highly specialized and engineering-heavy, requiring deep context, feedback loops, and human oversight. His core signal: AI is a general-purpose technology, but not that general — true value will come from those who learn to build with it and integrate human judgment, not from assuming AGI is imminent.
To go deeper, subscribe to my monthly Founder Insights newsletter, where I share lessons from the frontlines of company building, perspectives on AI’s future, and our industry’s road ahead.
