FMS 2025: Pliops Unlocks GenAI Inference with Smarter, Persistent Memory

Pliops is at FMS 2025: the Future of Memory and Storage to highlight its GenAI solutions – and demonstrate how they unlock scalable inference across agents, models and workflows.

Click to enlarge

With customers actively shipping Pliops FusIOnX-based solutions starting in Q3, the company is accelerating real-world adoption of smarter, memory-optimized GenAI systems.

At the core of GenAI’s intelligence lies its ability to remember – but today’s infrastructure wasn’t designed for the scale, context and complexity that GenAI demands. The company is shifting this paradigm with its XDP LightningAI with FusIOnX stack, which transforms inference by unlocking real-time memory reuse, enabling smarter collaboration and delivering scalable performance with seamless deployment. As memory bottlenecks continue to hinder GenAI’s growth, FusIOnX empowers systems to overcome these limitations, accelerating the path to more efficient and scalable AI.

“At Pliops, we believe memory is the missing piece in making GenAI truly intelligent and efficient,” said Ido Bukspan, CEO, Pliops. “Our FusIOnX goes beyond acceleration – it enables inference to remember, collaborate and evolve. It’s the infrastructure GenAI has been waiting for to power the next wave of scalable, memory-intensive LLM innovation.”

Whether building customer support agents, developer copilots, dialogue systems, or knowledge retrieval tools, the firm’s FusIOnX helps GenAI systems deliver smarter, faster and more resource-efficient results.

Built to Empower Smarter Collaboration
FusIOnX enables GenAI systems to think faster and respond more coherently by:

• Per-Agent KV-Cache Reuse for session continuity
• Shared Memory Tiers that support agent-to-agent logic
• Plug-and-Play Deployment into frameworks such as vLLM and NVIDIA Dynamo

Integration with GenAI Software
FusIOnX delivers plug-and-play compatibility with leading GenAI frameworks, while leveraging the high performance of LightningAI KV-store hardware to maximize persistent memory reuse. The result: accelerated inference, longer context retention, and reduced GPU load – without any need for retooling.

Pliops is partnering with Wiwynn Corp., a global cloud IT infrastructure provider, to co-develop advanced caching architectures that scale with GenAI workloads. “We are pleased to collaborate with Pliops to rethink memory for GenAI at scale,” said Ted Pang, senior director, Wiwynn. “By combining our strengths, we aim to enable smarter caching architectures designed to accelerate inference, conserve GPU resources, and unlock persistent intelligence across multi-agent systems.”

Redefining Memory for AI: Live at FMS
At FMS, Moshe Twitto, CTO and founder, Pliops will participate in a technical session focused on GenAI and the strategic integration of enhanced memory architectures to address the computational and memory challenges inherent to LLMs. This session will take place on Wednesday, August 6 from 8:30 – 9:35 a.m. PT. Additionally, Tony Afshary global VP, solutions and marketing, Pliops, will present a breakout session titled, Redefining Inference at Scale: New Memory Infrastructure for AI on Wednesday, August 6 from 12:55 – 1:10 p.m. PT.

Joining the team onsite will also be newly appointed Amit Golander, VP, products, further showcasing Pliops’ strategic investment in product innovation and leadership. Visit the company at booth #840 during FMS 2025, August 5-7 at the Santa Clara Convention Center, to see how FusIOnX is unlocking GenAI’s full potential.

To experience the benefits of smarter, persistent memory firsthand, an accelerated inference instance powered by Pliops FusIOnX is now available for trial at this link.

Stay tuned for major announcements at the upcoming AI Infra Summit, where Pliops will reveal new partnerships that further strengthen and expand the GenAI ecosystem.

Resources:
XDP LightningAI Solution Brief    
Blogs