
On-Demand Library
Every AI Leadership Forum session, with video and slides. Free to watch.
10 sessions
Episode 2 · Nov 5, 2026 · Market analysis
CXL Market Tear Down
Frank Berry, IT Brand Pulse
Frank Berry examines how AI demand, rising memory prices, hyperscaler investment and vendor strategy are reshaping the CXL market, from DDR4 reuse to a forecast $663 million merchant CXL memory-controller market in 2030.
Episode 2 · Nov 5, 2026 · KV cache
KV Cache Is a Cost Question
Jim Handy, Objective Analysis
Jim Handy weighs the cost of storing KV cache against recalculating it, and how to choose among HBM, system DRAM, CXL-attached memory and storage as memory prices change.
Episode 2 · Nov 5, 2026 · KV cache
Scaling AI with MemoryAI KV Cache Servers
Kumar Natarajan, Penguin Solutions
Kumar Natarajan explains how KV caching cuts repeated computation, and how the Penguin MemoryAI KV Cache Server uses up to 11 TB of combined host and CXL memory as a shared cache tier for enterprise inference.
Episode 2 · Nov 5, 2026 · Memory architecture
The Emerging Memory-Centric Data Center
Steve Scargall, Liqid
Steve Scargall shows how CXL memory expansion, pooling and sharing address growing KV cache demands, with benchmarks for throughput, time to first token and cross-server cache reuse, plus a buyer's checklist.
Episode 1 · Sep 30, 2026 · CXL fundamentals
CXL and the New AI Memory Hierarchy
Steve Scargall, Liqid
Steve Scargall, Chief Architect at Liqid, explains how CXL extends the AI memory hierarchy between local DRAM and NVMe storage, with benchmark results for CXL cache offload and cross-server KV cache sharing.
Episode 1 · Sep 30, 2026 · Memory architecture
The Imperative for Memory Innovation in the AI Era
Arthur Sainio and David Selle, Penguin Solutions
David Selle and Arthur Sainio explain how AI is reshaping the server memory hierarchy, where DDR5 RDIMMs, MRDIMMs, SOCAMM2 and CXL fit, and how CXL add-in cards give qualified DDR4 modules a second life.
Episode 1 · Sep 30, 2026 · Market analysis
Why CXL Matters
Frank Berry, IT Brand Pulse
Frank Berry opens the series with why memory capacity and cost have become two major challenges for AI infrastructure buyers, and how CXL helps through memory expansion, pooling and DDR4 reuse.
Episode 1 · Sep 30, 2026 · Market analysis
CXL Market Update
Jim Handy, Objective Analysis
Jim Handy covers what CXL does today (memory expansion, DDR4 reuse, pooling and shared memory), the DRAM price spike behind the interest, and his forecast of a CXL market of about $16 billion by 2031.
Episode 1 · Sep 30, 2026 · CXL fundamentals
CXL: From Standard to Industry Transformation
Anil Godbole, Intel
Anil Godbole of Intel, chair of the CXL Consortium Marketing Work Group, reviews the CXL 4.0 specification and the data center use cases gaining traction: memory expansion, DDR4 reuse, tiering and pooling.
Episode 1 · Sep 30, 2026 · Memory architecture
Big Data Feeds Big Compute Through Memory
Siamak Tavallaei, Samsung
Siamak Tavallaei of Samsung shares CXL research results, including a 40% latency reduction with hot-page monitoring and 50% more usable capacity through pooling, and proposes hybrid switches that bridge GPU interconnects and CXL memory.
About the library
The On-Demand Library collects every AI Leadership Forum session for AI professionals who plan, buy and build AI infrastructure. It holds 10 sessions from the first two episodes of Endless Memory with CXL, each with the full video and the speaker's slides.
What is the AI Leadership Forum?
The AI Leadership Forum is a webinar series produced by IT Brand Pulse. It brings together researchers, technology suppliers and end users to explain how AI infrastructure is being built.
Is the On-Demand Library free?
Yes. Every session is free to watch, and each speaker's slides are free to download with no sign-up.
What topics does the library cover?
The current series, Endless Memory with CXL, covers CXL fundamentals, KV cache for AI inference, memory architecture and market analysis. Episodes on DDR4 reuse and memory pooling air on December 2 and January 13.
Who presents at the Forum?
AI leaders from companies like Intel, Samsung, Liqid and Penguin Solutions, alongside industry analysts. New speakers join with each episode.
How do I watch an episode live?
Register free for an upcoming episode in the Coming up live section of this page. Each episode joins the library after it airs.
Coming up live
Two episodes remain in Endless Memory with CXL. Each joins the library after it airs.
Wednesday, December 2
Episode 3: DDR4 Reuse
How to identify, qualify and redeploy existing memory, and calculate what it saves.
Wednesday, January 13
Episode 4: Memory Pooling
How multiple hosts draw from a common memory resource, with a look at orchestration and deployment.
Want your session in the library?
Vendors sponsor a series. Researchers and end users speak free.