All articles tagged with #agentic ai
Nvidia CEO Jensen Huang says agentic AI—capable of researching, analyzing, and executing multi-step tasks—has pushed AI demand into a parabolic growth trajectory, signaling tangible business impact beyond simple Q&A; Nvidia’s stock has climbed, but investors should watch for potential deceleration as competition narrows margins and valuations remain high.
GF Securities analysts forecast NVIDIA will unveil Vera Rubin Arm‑based CPUs at Computex 2026, claiming Vera delivers up to 1.5x faster speed, 2x overall performance, and 4x higher density per rack versus AMD/Intel x86 chips for agentic AI workloads, with projected shipments of 1.2 million units in FY2027 and 4.2 million in FY2028. The move supports a growing CPU TAM for AI inference—potentially $211 billion by 2030 as agentic AI accounts for about 30% of inference computing—while Intel is expected to counter with Wildcat Lake, signaling a shift in the server hardware mix toward CPUs in the AI era.
Advanced Micro Devices (AMD) surged to a 52-week high with a year-to-date gain near 112%, as investors rotate out of Nvidia into AMD on faster growth and a smaller base. AMD’s strength in Agentic AI workloads boosts demand for its EPYC CPUs and Instinct GPUs, while MI355X accelerators gain traction for AI inference and ROCm’s maturation closes the CUDA gap. Price targets up to $625 imply meaningful upside despite a high trailing P/E, making AMD a leading momentum play for 2026 as the AI server market shifts in its favor.
Meta and Google are reportedly developing highly capable AI agents to perform tasks for users, joining the OpenClaw-driven wave of ‘agentic’ AI. Analysts say these agents could transform platforms into revenue engines through transactions, ads, or enhanced productivity, driving engagement and lock‑in, but security, governance, and trust remain major challenges as competition heats up among Big Tech, startups, and incumbents.
Google reportedly is testing Remy, a Gemini-powered AI assistant designed to act as a proactive digital operator inside the Gemini app, capable of managing work, school and personal tasks, learning user habits, and planning actions beyond simple questions. The move signals a shift toward agentic AI—integrating planning and task execution across Google services like Workspace, Android and Search—though no launch date is set and privacy concerns persist.
SAP plans to acquire Dremio to create an Apache Iceberg-native, open lakehouse that unifies SAP and non-SAP data, enabling real-time, AI-ready analytics and agentic AI with no data movement. The deal will bolster SAP Business Data Cloud and HANA Cloud, and is expected to close in Q3 2026 pending regulatory approvals.
CISA, with Australia’s ASD and other partners, released guidance for organizations adopting agentic AI, detailing key security challenges and risks and providing actionable steps to design, deploy, and operate these systems while aligning AI risk management with established cybersecurity frameworks and strengthening oversight as adoption grows.
Microsoft outlined a broad AI push: Agent Mode (now the default 'edit with Copilot') is integrated across Excel, Word, and PowerPoint, with Copilot and GitHub Copilot moving to usage-based pricing. Nadella said nearly 90% of Fortune 500 companies have active agents built with Microsoft’s low-code tools, and Copilot usage is rising as customers deploy custom agents. GitHub Copilot will include monthly AI credits, and Microsoft plans more AI infrastructure spending (~$40B). Bing reached 1 billion active monthly users, underscoring the company’s AI‑driven strategy.
Meta has partnered with Amazon Web Services to run agentic AI workloads on AWS Graviton Arm-based chips, aiming to scale and optimize Meta’s AI stack across its services with better performance and efficiency using AWS infrastructure.
Google Cloud announced a $750 million fund to accelerate agentic AI across its 120,000-strong partner ecosystem, offering tools, early access to Gemini models, credits for sandbox development, and embedded engineers with major partners to help prototype, deploy, and govern enterprise-ready AI agents for joint customers.
Adobe unveils an agentic Firefly assistant and partners with Anthropic to bring Claude-based capabilities to Creative Cloud, enabling autonomous image edits and cross‑app workflows, with a public beta arriving later this month.
Microsoft plans to turn Copilot into an always-on agentic AI that autonomously handles tasks across apps (email, calendar, to-dos) using OpenClaw-like tech with safety guardrails, ahead of Build 2026.
Autonomous AI agents capable of thinking, acting, and adapting without human intervention are being described as a hacker’s dream, potentially enabling scalable, relentless cyberattacks on corporate and government systems. Fortune highlights Anthropic’s Mythos as far ahead in cyber capabilities, signaling a new wave of AI-driven exploitation, while ‘shadow AI’ risks grow as employees run agents remotely. A Dark Reading poll shows nearly half of cybersecurity professionals view agentic AI as the top attack vector for 2026, underscoring the urgent need for safe, controlled AI environments and heightened organizational awareness.
Samsung Electronics announced Samsung Browser for Windows, extending its mobile browser to PC with cross-device continuity and a new agentic AI assistant built with Perplexity. The browser lets users pick up where they left off across devices, uses Samsung Pass for autofill, and employs natural-language understanding to manage tabs, summarize content across multiple tabs, search history by natural language, and even locate moments in video content. Availability covers Windows 10/11 (version 1809+), with agentic AI features initially in South Korea and the United States and plans to expand to more markets; supported devices include Galaxy Book3–6 series and required Samsung Account and Continuity Service.
Anthropic’s Claude now includes agentic AI that can remotely operate a user’s macOS computer to perform repetitive tasks—like batch-resizing photos in Photoshop and adding logos—via the Claude Dispatch app, with previews limited to Claude Pro/Max; it prioritizes connectors to services but can execute tasks directly if needed, raising security concerns about accessing sensitive files.