Lenovo Servers Use AMD EPYC and NVIDIA AI Services

Key Takeaways

• Rick Wurster says Charles Schwab blocks unauthorized AI agents while building controlled external account connections; banks are weighing fraud, liability and security risks.
• A 2026 survey found 29% of 2,400 knowledge workers reported actively sabotaging their employer’s AI strategy, rising to 44% among Gen Z.
• Lenovo’s ThinkSystem SC755 V5 uses liquid cooling to remove up to 100% of system heat and deliver up to 40% higher sustained performance than similar air-cooled systems.
• Lenovo’s AI infrastructure uses 6th Generation AMD EPYC SP7 processors and includes expanded services with NVIDIA Infrastructure Services expertise.
• ASUS’s Agent Computer initiative supports local, cloud and hybrid AI on systems using NVIDIA RTX or AMD Radeon AI GPUs; its mini PC supports workflows with up to 120-billion-parameter models.
• AWS recommends its six-stage Agent DLC process—Define, Build, Evaluate, Release, Observe and Feed Back—and starting with one focused task and an evaluation baseline within a week.
• Harvard’s David Parkes sees potential in societies of specialized AI agents, while identifying scalability, privacy, reliability and trustworthy agreements as challenges.
• Khalifa University and Harvard researchers report 90% accuracy for an AI health-risk system using genetic, medical-history and lifestyle data; it is designed to support doctors, not replace them.
• The California Federation of Teachers seeks a three-year pause on AI in public K–8 classrooms, citing readiness, student development and privacy concerns, including research involving ChatGPT and GPT-4.
• An editorial says an AI-assisted app could enhance the Gibbes Museum of Art’s 3D exhibits for visitors with limited vision.

AI agents could weaken businesses built on customer inertia

Companies often rely on customer habits, switching hassles, and hard to compare prices to protect their market share. The article calls these advantages behavioral moats and says personal AI agents could weaken them by finding better deals and helping customers switch. Financial services and insurance are examples, since comparing accounts or policies can take time. Schwab CEO Rick Wurster said Schwab blocks unauthorized AI agents while building controlled ways for outside systems to connect. Security, trust, and how much control people will give agents remain open concerns.

Banks should block outside AI agents while building safeguards

Banks have reason to move carefully as customers begin using outside AI agents, but they may not be able to block them forever. AI agents could make it easier for people to compare offers and move money, while creating new fraud and liability risks. Charles Schwab is blocking unauthorized agents and developing controlled account connections, while six banks, including Bank of America and Capital One, are also working on the issue. Approved connections still need strong protections, since agents could create risks that current systems may not detect. The article says banks should prepare policies and improve coordination among fraud, cybersecurity, and compliance teams.

Leaders can ease AI fears by talking with workers

Workers may value generative AI while also fearing what it could mean for their jobs and daily lives. The article says this uncertainty can make employees resist workplace changes, so leaders should discuss those concerns rather than push ahead without conversation. A 2026 survey of 2,400 knowledge workers in the U.S., U.K., and Europe, including 1,200 executives, found that 29% of employees said they had actively sabotaged their company’s AI strategy. That figure rose to 44% among Gen Z workers. The article argues that calls to slow down AI are a reason to pause and talk, not to stop AI entirely.

Lenovo adds liquid cooled servers and services for AI workloads

Lenovo expanded its infrastructure portfolio for AI and high performance computing, including ThinkSystem servers with 6th Generation AMD EPYC SP7 processors, Neptune liquid cooling, and a colocation partnership with Digital Realty. The ThinkSystem SC755 V5 can remove up to 100% of system heat with liquid cooling and deliver up to 40% higher sustained performance than similar air cooled systems. Lenovo also introduced the ThinkSystem SR755, storage arrays DE6800 and DE6500, and validated the Cornelis CN5000 interconnect. Its expanded Hybrid AI Factory Services include GPU deployment, configuration, and optimization, with NVIDIA Infrastructure Services expertise supporting specialized deployment and cluster validation.

ASUS brings local AI agents to desktops with new ecosystem

ASUS introduced its Agent Computer initiative to bring local and agentic AI to DIY PCs, gaming systems, mini PCs, and workstations. Users can choose local, cloud, or hybrid AI based on performance, privacy, and cost needs, using systems built with ASUS motherboards and NVIDIA RTX or AMD Radeon AI GPUs. ASUS also launched an online tool called Find Your ASUS Agent Computer to suggest systems or components. Its software includes Zenni Claw for agentic AI and Multi-LM Tuner for deploying optimized large language models. The company says its mini PC can run local AI models and workflows with up to 120 billion parameters.

AWS outlines a path for taking AI agents into production

AWS describes Agent DLC, a process for moving AI agents from successful demos to production. It links six activities: Define, Build, Evaluate, Release, Observe, and Feed Back. Teams should set clear business goals and test agents across measures such as accuracy, cost, speed, capacity, and availability. Because agents and the services they rely on can change, teams should record execution traces and review real-world interactions to find problems. AWS recommends starting with one focused business task and building an evaluation baseline within a week.

AI researchers envision societies of specialized agents

David Parkes, a Harvard computer science professor and recipient of the IJCAI-2026 John McCarthy Award, discussed the future of AI agents. He imagines personalized agents that learn people’s interests, priorities, and goals, but says scalability and privacy remain major challenges. Multi-agent systems are already used in software development, though Parkes says their design is still largely based on trial and error. He describes agent societies as millions of AI modules with different skills and goals, potentially working together on scientific research. Such systems will need ways to check agents’ reliability and create trustworthy agreements, since agents could make mistakes or reveal private information.

California teachers call for three-year pause on AI in K-8 schools

The California Federation of Teachers urged Gov. Gavin Newsom to pause AI use in public school classrooms from kindergarten through eighth grade for three years. The union, which represents 120,000 educators and faculty, says schools lack safeguards and teachers are not ready to supervise the tools. Its letter warns that AI could harm student development, reduce teachers’ control over instruction, and expose student data. It cited an MIT study linking ChatGPT use to weaker brain connectivity during essay writing and research finding that students who used GPT-4 scored 17% lower on a later unassisted math test. The union asked Newsom to issue an executive order before his term ends in January.

Researchers develop AI system to estimate future health risks

Scientists at Khalifa University in Abu Dhabi and Harvard University developed an AI system that analyzes genetic data, medical history, and lifestyle to estimate the chance of developing certain diseases. Presented at the World Economic Forum in Davos, the system is said by the team to predict health outcomes with 90% accuracy. Khalifa University CEO Dr. Ahmed Al-Khamisi said the goal is to make healthcare more proactive, while stressing that the system is meant to support doctors rather than replace them. Researchers are refining the system, planning more studies, and working on a mobile app for health tracking and personalized recommendations. They also aim to develop guidelines for AI use in healthcare that is transparent, explainable, and fair.

Editorial weighs AI’s promise and its risks

The Post and Courier editorial page says its editorials represent the newspaper’s institutional view and are produced separately from the news department. The article notes that Charleston’s Gibbes Museum of Art has 3D exhibits for visitors with limited vision. It says a new AI-assisted app could make that experience more engaging.

Sources

• Which business moats will still matter in the age of AI?
• Blocking AI agents is the right near-term move for banks
• Want to drive AI innovation? Talk about feelings
• Lenovo Launches Products for Optimising AI and HPC Workloads
• ASUS Aims To Turn Desktops Into A Local AI “Agent Computer” With New Hardware And Software Ecosystem
• From prototype to production: Engineering agents with Agent DLC
• From Intelligent Agents to Agent Societies
• California teachers urge 3-year pause on AI use in K-8 schools
• Can AI predict the future of your health?
• Editorial: The potential peril of AI is tremendous. So is the potential promise.