Large Language Models

Today's Stories

1. 1

   Micron Poised for AI Memory Boom as Agentic Systems Demand Ramps

   Micron reported third-quarter fiscal 2026 revenues of a record $41.46 billion, up 345.7% year over year and 73.7% sequentially, driven by AI demand. The company expects industry data center DRAM and NAND bit shipments in 2026 to more than double from two years earlier, as agentic AI workloads—systems that reason, plan, and execute tasks—require more memory across accelerator racks, CPU racks, and storage. Agentic AI expands memory needs beyond accelerator chips alone into CPU and storage infrastructure, potentially making memory a larger share of overall AI infrastructure spending. Micron's Core Data Center Business unit revenues jumped to $11.52 billion from $1.53 billion year-over-year, signaling strong near-term capture of this shift.

   Micron's HBM4 is in high-volume shipment and HBM4E is expected to enter production in 2027; the company's PCIe Gen6 SSD is purpose-built for agentic workloads. Zacks consensus estimates project Micron fiscal 2026 revenues of $129.61 billion (246.8% year-over-year growth) and fiscal 2027 revenues of $248.08 billion (91.4% growth).

2. 2

   Citi analyst notes Baidu's AI compute opportunity, ERNIE focus amid LLM competition

   Citi analyst Alicia Yap discussed Baidu's latest earnings, highlighting the company's renewed focus on AI models and its position in China's competitive large language model (LLM) environment. Baidu operates in a highly-competitive LLM market in China, and the analyst's commentary on the company's strategy—particularly around compute infrastructure and its ERNIE AI model—provides insight into how the company is positioning itself against rivals in the sector.

   The discussion centers on two areas: the near-term opportunity in compute resources and Baidu's longer-term focus on ERNIE model development, as well as performance of its Apollo Go autonomous vehicle business, which the analyst highlighted as a standout.

3. 3

   Microsoft pushes AI deeper into schools with governance, context controls

   Microsoft 365 Education is introducing agents and Copilot experiences designed to work within existing institutional systems—helping educators streamline lesson planning, assessment, and differentiation, and enabling student-support staff to coordinate aid and advising without repeated handoffs. The shift moves AI from isolated experimentation to institution-wide use tied to real workflows. According to Microsoft's 2026 AI in Education report, 92% of surveyed students and education leaders and 88% of surveyed educators are already using AI for school purposes. However, general-purpose AI lacks the institutional context and privacy/integrity safeguards schools require. Microsoft 365 Education addresses this by grounding AI in each school's identity, permissions, compliance, and data governance—keeping institutional data under the school's control while limiting access to what users are authorized to see.

   Institutions can start with Copilot Chat (included in Microsoft 365) and expand to richer Copilot experiences and agents as governance and readiness mature. Model choice depends on the task and the school's data-processing requirements, not just which model is newest. The approach lets schools pilot a single high-impact workflow before committing institution-wide.

4. 4

   Aily Labs Brings AI Decision Platform to Google Cloud Marketplace for Pharma

   Aily Labs, an AI Decision Intelligence platform company, partnered with Google Cloud to make its platform available on the Google Cloud Marketplace. The move lets pharmaceutical and biotech firms buy Aily's AI agents directly through existing Google Cloud contracts, and those purchases count toward their committed cloud spend without triggering separate vendor reviews or new security evaluations. Pharma and biotech companies typically juggle R&D, clinical trials, regulatory compliance, and supply chains across dozens of disconnected systems—delays in trial site selection, drug candidate prioritization, and inventory decisions often result. By offering Aily's platform through Google Cloud, enterprises can tap AI agents for drug development and regulatory work without extra procurement friction or security audits.

   Aily's "Super Agent" orchestrates thousands of specialized AI sub-agents across R&D, clinical development, regulatory filings, capital allocation, supply chain, and commercial operations—all accessible via the Agent Gallery within Gemini Enterprise, Google Cloud's enterprise AI application.

5. 5

   Salesforce charts layered data architecture for enterprise AI

   Salesforce's head of data architecture, Tristan Baker, outlined how knowledge graphs are becoming a foundational layer for enterprise AI systems. He described a layered stack where lakehouses and operational databases sit below a metadata layer, with graph technology managing relationships and context across them—enabling AI agents to reason from trusted data rather than simply retrieve it. Enterprise leaders are seeking conversational interfaces that return trustworthy answers in seconds, but that requires far more than a single database. Baker emphasized that the hard problem is accurately describing enterprise information so AI can make sense of it across fragmented data landscapes. Governance and semantic access policies are emerging as critical challenges; legal teams need to define data protection at the semantic level, not just within individual databases.

   Baker noted that most organizations have not yet fully solved this architecture problem. He cautioned that even the best technology yields "garbage in, garbage out" if the content exposed to it is not carefully curated—a challenge many enterprises are still grappling with.

6. 6

   Mizuho Financial Group deploys on-premise GPUs instead of cloud

   Mizuho Financial Group has introduced on-premise GPUs (eight NVIDIA units) rather than relying on cloud infrastructure, positioning itself to use AI agents. On-premise deployment gives Mizuho direct control over its AI infrastructure and data, which may be critical for a financial institution managing sensitive customer information and seeking to reduce dependency on external cloud providers.

   The company is preparing to deploy AI agents — autonomous systems that can perform tasks independently — suggesting a shift toward more sophisticated automation in financial operations.