Software operates in controlled environments. Robots do not.
On a recent visit to @Rhoda AI, I saw its robots in action.
As Rhoda founder @Jagdeep Singh told me, “You can’t script the real world.” So I tested how the robots would respond. I introduced an unexpected object, repositioned a radar during a precision task, and placed obstacles in a robot’s path.
The robots adjusted, recovered, and kept working.
Robots operating in factories, warehouses, and other customer environments cannot depend on perfectly controlled conditions. They must recognize change, adapt in real time, and complete the job reliably.
Rhoda trains its models on internet-scale video, then post-trains them with a smaller amount of real-world robot data. This helps the robots respond to situations they weren’t explicitly programmed to encounter.
That adaptability is essential for robots to operate reliably in factories, warehouses, and other customer environments.
Jagdeep and the Rhoda team are building the full stack, from AI models to hardware, and applying it to real customer use cases, including one for a major German automotive manufacturer.
In the news this week, Analog Devices agreed to acquire Alif Semiconductor for $1.35 billion to expand on-device AI, and Qualcomm and Amazon struck a multi-generation custom AI silicon partnership, while NVIDIA and its Australian data-center partners planned up to 2 gigawatts of capacity by 2027. Oracle brought 850 megawatts online in a single quarter as cloud infrastructure revenue reached $7.4 billion, and Mistral raised €3 billion at a valuation above €21 billion.
Take a walk with Jagdeep and me to see what Physical AI is already making possible: https://lnkd.in/p/geHj6YyF
Bottom line: Physical AI will usher in the era of Labor as a Service, transforming how work gets done across a global labor market worth more than $30 trillion.
In the news this week, Analog Devices agreed to acquire Mayfield portfolio company Alif Semiconductor for over $1.5 billion to expand on-device AI, and Qualcomm and Amazon struck a multi-generation custom AI silicon partnership, while NVIDIA and its Australian data-center partners planned up to 2 gigawatts of capacity by 2027. Oracle brought 850 megawatts online in a single quarter as cloud infrastructure revenue reached $7.4 billion, and Mistral raised €3 billion at a valuation above €21 billion.
Full Weekend Edition below. 👇
Signals Shaping the Future of AI:
Infrastructure
NVIDIA and Australian data-center partners plan up to 2 gigawatts of AI infrastructure by 2027. Firmus, Sharon AI, IREN, ResetData, CDC, NEXTDC, and AirTrunk will build capacity for NVIDIA DSX AI factories. Click here
Qualcomm and Amazon strike a multi-generation AI silicon deal tied to up to $60 billion in potential purchases. Qualcomm will build custom AI inference chips and optical networking technology for AWS, while Amazon receives warrants for up to 25 million Qualcomm shares. Click here
Oracle brings 850 megawatts of additional data-center capacity online in one quarter. Oracle Cloud Infrastructure revenue reached $7.4 billion, up 121% year over year, as AI-driven cloud demand accelerated. Click here
HP, Red Hat, and NVIDIA extend production AI infrastructure from data centers to the edge. The planned platform combines HP ZGX Fury, NVIDIA GB300 Grace Blackwell Ultra, and Red Hat AI Factory for local, cloud, and hybrid inference. Click here
Soluna completes an 83-megawatt Texas site and secures conditional base-load status for its full 166-megawatt campus. Kati 2 is planned to add 83 megawatts for AI and high-performance computing. Click here
Enterprise
Meta launched Muse, a personal AI agent that can access email, calendars, payment services, and other apps to complete tasks autonomously. The rollout highlights how agent competition is shifting beyond model quality toward distribution, personal context, and secure access to sensitive workflows. Click here
OpenAI launches ChatGPT for Financial Services with Morgan Stanley and Evercore. The tailored ChatGPT Work offering combines GPT-6 Astra with built-in data from PitchBook, LSEG News, Daloopa, and Crunchbase for research, modeling, and client work. Click here
Salesforce introduces an enterprise AI Control Plane for agents across Salesforce and third-party systems. The Trusted Enterprise AI Harness adds discovery, identity, policy, lifecycle, evaluation, observability, and cost controls across enterprise AI. Click here
Wipro says AI productivity gains freed capacity equivalent to 20,000 employees. The company redeployed that capacity and is shifting toward a human-AI operating model, with more than 100,000 employees receiving advanced AI training. Click here
Cisco, Palantir, and NVIDIA combine sovereign AI software with a controlled enterprise infrastructure stack. Cisco Secure AI Factory with NVIDIA becomes a preferred full-stack foundation for Palantir Sovereign AI OS and its cybersecurity ontology. Click here
Accenture and Google Cloud launch a 1,000-engineer group for enterprise agent deployments. The Gemini Enterprise Business Group will embed forward-deployed engineers with clients to move AI from pilots into scaled operating workflows. Click here
Capital Flows
Analog Devices agreed to acquire Alif Semiconductor for over $1.5 billion to expand its on-device AI capabilities. The deal combines ADI’s sensing and power-management technologies with Alif’s low-power AI processors for real-time consumer and industrial applications. Click here
Harvey raised $550 million at a $15.5 billion valuation as adoption of its legal AI platform continues to expand. The company says 80% of Am Law 100 firms now use Harvey, and the new funding will support further product development, hiring, and expansion across law firms and in-house legal teams. Click here
Mistral raises €3 billion at a post-money valuation above €21 billion. Samsung Electronics led the Series D, the largest equity round completed by a European technology company, to expand model research, compute capacity, and infrastructure. Click here
Cognition raises $2 billion at a $48 billion valuation. Andreessen Horowitz and Accel led the Series E as the developer of the autonomous software-engineering agent Devin nearly doubled its valuation from May. Click here
Zankore signs up to $3.1 billion in senior debt financing for NVIDIA AI infrastructure. The facility will fund GPU and cloud computing deployment as the Ooredoo-backed platform expands AI capacity across Indonesia and Southeast Asia. Click here
Positron raises $875 million at a $5 billion valuation for AI inference chips. The round will fund its Asimov processor and Titan systems as investors back alternatives to incumbent AI accelerators. Click here
Research
OpenAI says an internal AI system used roughly 10,000 coordinating agents over 88 hours to produce a proposed solution to the Navier-Stokes Millennium Prize problem. The effort generated 2.7 million agent messages and 130 billion output tokens, highlighting the growing role of large-scale agent orchestration in scientific discovery. Click here
DeepSeek releases V4.1-Flash with native vision and a sharply smaller inference memory footprint. Its KV cache uses one-quarter the HBM and one-eighth the SSD storage of the previous generation, alongside lower API pricing. Click here
Google DeepMind releases molecular-effect predictions for all 9 billion possible single-letter human DNA variants. AlphaGenome Atlas makes the catalog available to academic researchers through a free web portal. Click here
Google and NASA JPL introduce an AI model that detects 50% more methane plumes than human experts. MAPL-EMIT identified more than 23,000 additional plumes globally using NASA’s EMIT satellite instrument. Click here
Peer-reviewed research finds that language models causally use confidence signals to guide behavior. Nature Machine Intelligence reports interventions showing model confidence affects whether systems answer or abstain, providing evidence of functional metacognition in language models. Click here
Policy
California enacts first-in-the-nation standards for independent AI audits and assessments. Senate Bill 813 establishes a framework for independent verification organizations, while Assembly Bill 1405 establishes an AI auditor registry and independence standards. Click here
California signs new safety requirements for AI chatbots used by minors. The measures require safeguards and risk assessments for chatbot operators and expand state oversight of AI systems affecting children. Click here
The EU places ChatGPT, Reddit, and Roblox under heightened Digital Services Act obligations. ChatGPT was designated a Very Large Online Search Engine, while Reddit and Roblox were designated Very Large Online Platforms. Click here
Bipartisan House lawmakers introduce the Stop Rogue AI Act. The bill would direct NIST to develop standards for identifying and stopping unauthorized AI agents operating inside business and federal networks. Click here
Global AI Strategy
U.S. security agencies formally warn that Chinese AI companies are conducting industrial-scale model distillation. NSA, FBI, and CISA issued a joint advisory describing systematic extraction of restricted capabilities from U.S. frontier models. Click here
The Pentagon is in talks to provide AI cloud provider Fluidstack with a $5 billion loan. The financing would be the largest ever from its Office of Strategic Capital and would support U.S. AI data-center supply chains and manufacturing capacity. Click here
Canada launches a National AI Literacy Initiative as part of its AI for All strategy. The program targets up to one million post-secondary students and more than 50,000 K-12 educators while expanding national workforce AI training. Click here
Abu Dhabi’s G42 is exploring U.S. majority ownership to preserve access to advanced AI chips. The talks show export controls beginning to shape not only access to compute, but the ownership structure of national AI champions. Click here
South Korea launches a presidential National AI Strategy Committee as its top-level AI policy body. The committee will coordinate national programs, monitor implementation, and set the Korea AI Action Plan across three policy pillars and twelve strategic domains. Click here
The UK publishes a government AI Risk Management Toolkit for AI design, procurement, and delivery. The framework provides a common process for identifying, quantifying, and treating AI risks across public and private-sector projects. Click here
Talent Signals
Each week, we spotlight key roles tied to the themes shaping this week’s AI headlines, connecting talent to the companies driving the news.
RhodaAI is building general-purpose robotic intelligence for industrial environments, using internet-scale video and real-world data to train robots that can adapt across complex physical tasks. Following its $450M funding round earlier this year, the Palo Alto-based company is expanding across robotics, AI, hardware, and engineering roles. Click here
ChaiDiscoveryis building AI models and software for molecular design, helping scientists use AI in high-stakes drug discovery and biological research workflows. Following its recent $400M financing, Chai is hiring across research, engineering, platform, security, and partnerships roles. Click here
CodeRabbit is building an independent code-review layer for AI-generated software, helping developers and coding agents ship faster without lowering quality or accountability. Following its recent $143M funding round, CodeRabbit is hiring across engineering, AI, product, and go-to-market roles. Click here
You can see all the opportunities at Mayfield-backed AI companies 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.
Nikesh Arora (Click here) — “Not using AI could and will become an existential issue for both individual users and enterprises. Employees and developers will take matters into their own hands since they will find their cautious enterprises aren’t moving fast enough. These activities will cause adoption sans security.” In a post viewed 1.3M times, Arora argues that AI adoption and AI security must advance together. As employees deploy coding tools, agents, and enterprise models faster than governance can keep up, he sees protecting IP, securing agent lifecycles, and maintaining real-time controls as becoming core requirements for enterprise AI deployment.
Jacob Coxon (Click here) — “I resigned from Anthropic today. I spent the last three years doing pre-training research at both OpenAI and Anthropic. Neither company is acting responsibly. They are racing straight to self-improving superintelligence and gambling with our lives.” In a post viewed 164.9M times, Coxon argues that frontier labs are moving too quickly toward increasingly autonomous and self-improving systems without sufficient safeguards or coordination. His departure turned into one of the week’s biggest AI debates, raising questions about whether competitive pressure between labs is outpacing the governance needed for the next generation of models.
Yoshua Bengio (Click here) — Turing Award-winning AI pioneer Yoshua Bengio writes, “We don’t know with certainty what comes next, but we know where these issues originate, and this can help us plan the path forward.” Following recent incidents in which advanced AI agents lied, cheated, coordinated, and breached infrastructure, Bengio traces these behaviors to deeper incentives created by reinforcement learning and goal-seeking systems. His argument is that as models become more capable, these behaviors could become more severe unless labs rethink how agents are trained and strengthen monitoring and governance.
