Welcome to
the Forum

Equinix’s Glenn Dekhayser and Solidigm’sScott Shadley discuss how power, cooling, and cost considerations are causingenterprises to embrace co-location among their AI infrastructure strategies.

We sat down with Dr. Cliff Federspiel, founder, president & CTO at Vigilent, to talk about compute efficiency in the AI era—specifically, how smarter cooling optimization can keep SLAs tight as rack densities rise. From sensing to predictive control, Cliff explains why letting an on-prem AI continuously tune the cooling plant beats static setpoints, helping operators protect reliability while cutting energy and carbon.

The company builds an on-prem, vendor-agnostic AI control layer for data-center cooling. A sensor network and machine learning generate an Influence Map® of how each AHU affects rack inlet temperatures; the system then adjusts fans and unit states only where needed. Vigilent integrates with BMS/DCIM and also monitors and controls chillers, which allows for optimization across the entire cooling plant. Guardrails and fail-safes—including snap-to-full-cooling—ensure resilience while delivering measurable efficiency gains.

Q1: Vigilent has been applying AI/ML to cooling optimization for 17 years—well before today’s wave of AI-driven data center demand. How has this early market entry shaped your leadership position as operators now face power densities and cooling challenges beyond the limits of traditional air cooling?

A1: Data centers are highly variable in how they’re configured and operated. There are different cooling technologies from different vendors. Control strategies also vary. For example, temperature vs. pressure. Facility layouts are different, for example, raised floors vs. slab and varying types of containment. As you say, Vigilent has been around for a long time, and that means we’ve seen all of these configurations and learned how to optimize cooling in each case. Years ago, we scratched our heads when seeing something new, but by now we’ve pretty much seen it all, including the complexities associated with optimizing across the air-side and chiller plant.

Increasing IT power densities just add another layer of complexity to the cooling challenge, and we’ve got the experience and gray hairs, and software and engineering talent, needed to deal with it.

Q2: Most traditional cooling systems rely on static setpoints, while Vigilent uses machine learning to understand the influence of each cooling unit on IT equipment across a facility. How does this dynamic approach transform cooling efficiency compared to conventional methods, and what measurable outcomes—like SLA compliance, energy use, or equipment longevity—can operators expect?

A2: Machine learning enables Vigilent’s AI to empirically understand exactly what’s going on in a data center. If a fan is ramped up or down, or a cooling unit is turned on or off, where will temperatures go up, and where will they go down? Machine learning allows the AI to create a predictive model, basically knowing in advance the effects of any actions it takes. This enables the AI to deliver very high SLA compliance with a bunch of other benefits. One colocation operator went from about 94% SLA compliance to 99.96% compliance. In the same data hall, this operator reduced PUE by 10% and energy and carbon emissions by 32%. Since cooling is only used when and where it’s needed, there has been a big reduction in wear and tear, which means their cooling infrastructure will last longer and there are fewer replacement parts. They are now using the AI as a competitive differentiator vs. other colocation operators.

Q3: With over 1,000 deployments across 35 countries and partnerships with companies like Schneider Electric, NTT, and Siemens, how does Vigilent’s global footprint enhance your AI models, and what unique advantages does this breadth of operational data provide to new customers, particularly those entering high-density AI computing environments?

A3: To address the diversity in data center designs and operations, Vigilent has developed platform capabilities that complement our core AI technology. We integrate with any type of cooling infrastructure, whether it be air cooling, liquid cooling, or the chiller plant, and also with BMS and DCIM systems plus other assets like power equipment. We also rapidly deliver bespoke capabilities with a tool called Vigilent Studio. And we have an information layer in our platform called Vigilent Insights, which uses the data captured and generated by Vigilent’s AI to provide facility staff with guidance about how to improve resilience or operate more efficiently.

Our strategic partners are global leaders in providing infrastructure and services to data center operators. This has positioned us well for the increased densities we’re seeing now. For example, Schneider Electric acquired the liquid cooling company Motivair and collaborates with NVIDIA on designs for AI data centers.

Q4: As rack densities push past 100 kW and liquid or hybrid cooling strategies become mainstream, where do you see AI-driven optimization playing the biggest role? How can it accelerate both sustainability and carbon reduction goals for next-generation infrastructure?

A4: Joint optimization of the air-side and water-side of the cooling plant will be increasingly important as rack densities rise and liquid cooling is used to remove some, but in most cases, not all, of the IT heat load. Higher densities shorten ride-through times in the event of a cooling failure, and hybrid cooling increases the complexity of the cooling plant. There is plenty of research showing that AI-driven vehicles are safer and more fuel-efficient than human-driven vehicles. Similarly, AI-driven optimization not only improves the efficiency and sustainability of complex data centers, but it also improves their reliability.

Q5: For operators considering AI-driven cooling optimization, what are the top three factors they should prioritize in evaluating solutions? And where can readers learn more about Vigilent’s approach to dynamic cooling management?

A5: People can be understandably nervous about letting AI control data center cooling, just as people are nervous getting into a self-driving car. But with proper safeguards, they can ensure data center resilience and be rewarded with significant benefits. What are those safeguards?

• First, make sure the AI operates within the premises, within the corporate firewall. This will avoid security risks associated with cloud-based applications.

• Second, make sure the AI has guardrails that protect against hallucinations, and fail-safes that ensure full cooling if there is ever a problem with the AI.

• Third, make sure the AI is proven. Has it been deployed in other mission-critical environments? How many? What were the results?

Learn more about Vigilent.

CEO Lisa Spelman explains how tackling hidden inefficiencies in AI infrastructure can drive enterprise adoption, boost performance, and spark a new wave of innovation. Check out Cornelis Networks.

This week, electronic design automation (EDA) leader Synopsys unveiled expansions to its Synopsys.ai Copilot generative AI capabilities, promising to transform workflows for semiconductor engineering teams, enabling them to take on more complex designs on accelerated timelines.

Synopsys reports that the new generative AI (Gen AI) capabilities accelerate workflows that previously consumed days into tasks completed in mere hours, and processes that took hours into operations finished in minutes. Early adopters are already reporting extraordinary gains: 30% faster ramp time for early-career engineers using knowledge assistant and 35% boost in engineering productivity for early-career and expert engineers within formal verification workflows using formal assertion assistant.

At the heart of this transformation lies Synopsys.ai Copilot’s dual approach: assistive and creative AI capabilities. The assistive features focus on knowledge management and workflow optimization, helping engineers to navigate complex documentation, generate scripts, and guide newcomers through the labyrinthine world of chip design. The creative capabilities venture into more ambitious territory, automatically generating formal assertions and register-transfer level (RTL) code with greater than 80% syntax accuracy and more than 70% functional accuracy.

This AI expansion gains additional significance when viewed through the lens of Synopsys’ recently completed acquisition of Ansys. The integration is already paying off with the recent introduction of Ansys Engineering Copilot and updates to Ansys SimAI, extending AI capabilities deep into simulation and analysis workflows.

In addition, Synopsys is looking beyond current generative AI applications toward what it calls AgentEngineer technology for chip design. Developed in collaboration with Microsoft, this represents a progression toward increasingly autonomous design systems. The company envisions a five-level evolution: from step-level actions (L2) to complex multi-agent operations (L3), dynamic flow optimization with adaptive learning (L4), and ultimately autonomous decision making (L5).

The first prototype offers a glimpse of a future where AI agents don’t just assist engineers but actively participate in the design process. As Microsoft’s Aseem Datar notes, “Together, we are not just optimizing existing workflows—we are introducing a new paradigm to advance engineering innovation and productivity for next-generation chip designs.”

The TechArena Take

In an industry where time-to-market can determine billion-dollar market positions, Synopsys has just handed its customers a significant advantage. The timing is particularly crucial as the industry grapples with a combination of design complexity and workforce shortages. Advanced AI chips require sophisticated design methodologies, while simultaneously, the pool of experienced semiconductor engineers remains constrained.

Synopsys’ AI capabilities directly address this paradox by accelerating novice engineers’ learning curves while amplifying expert productivity. The real test will be execution at scale, but the early results suggest meaningful progress toward AI becoming a more integral part of the semiconductor design process.

‍

Explore myths, metrics, and strategies shaping the future of energy-efficient data centers with Solidigm’s Scott Shadley, from smarter drives to sustainability-ready architectures.

Equinix’s Glenn Dekhayser and Solidigm’s Scott Shadley join TechArena to unpack hybrid multicloud, AI-driven workloads, and what defines a resilient, data-centric data center strategy.

As AI drives power demands sky-high, hyperscale leaders share opportunities, obstacles, and the urgent path forward for immersion cooling adoption.

‍

Alistair Bradbrook, founder and COO of Antillion, has spent decades wrestling with one central challenge: getting the right information to the right people at the right time—no matter the environment. From early exposure to Phase I clinical trials at his father’s pharmaceutical company to leading edge-AI hardware design, Bradbrook’s career is a study in curiosity, iteration, and purposeful design.

In this conversation, he shares how Antillion approaches technology, leadership, and research/testing at the edge.

Q: You helped found Antillion with a mission to bring cutting-edge performance to the edge. What was the spark that told you this company needed to exist?

A: There wasn’t one epiphany—it was a series of moments over decades. My interest in collecting data at the source started when I was 12 or 13. I worked with my father in his business, which was a pharmaceutical company that did Phase I trials, testing drugs in humans for the first time. The challenge in that business had always been about collecting data—like heart rate, ECG, and all those metrics that are key to understanding the effect of drugs on humans—quickly enough to aggregate them so decisions could be made.

This was ~35 years ago, so the idea of putting high-performance computing at the edge was unimaginable. We were working with the home computers of the day—not exactly capable edge devices. But the challenge was the same as today: collect data close to its source, aggregate it quickly, and empower people to make decisions in real time.

That fascination stayed with me—whether in healthcare, environmental sensing, or defense—and eventually led me to Antillion. I’d grown frustrated with existing products and felt that if no one else was building the kind of capable, usable, and non-intimidating technology I envisioned, then we should do it ourselves.

Q: Antillion’s products blend engineering rigor with industrial design aesthetics. What does “good design” mean to you in the context of edge technology?

A: My industrial design lead and I start pretty much conversation with a few factors that are really critical. The object we’re building needs to not feel alien—it needs to not feel so brand new that it’s not relatable and people don’t understand how to use it. So, coming up with designs that are not alienating is key.

We also think a lot about portability—not just size, weight, and power (SWaP) in the defense sense, but true portability. Can one person move and operate it easily? Does its form factor match its purpose? We’ve built devices that, in hindsight, didn’t meet that standard—they were too big for their capability or too small to be useful. That balance matters.

We also iterate quickly, moving from sketch to 3D design to a first physical prototype, because some flaws only reveal themselves when you can hold and use the object. Perfection is unattainable, but each release gets closer.

Q: Running a fast-moving company at the frontier of hardware and edge AI is no small feat. What have you learned about leadership in unpredictable environments?

A: My style is collaborative and passion-driven. I’ll never ask someone to do something I wouldn’t do myself. In a ~30-person company, I can stay connected with everyone, but as we grow, that culture will rely on managers carrying it forward.

I want people who have genuine interest in what we’re building—not just a nine-to-five mindset. The best moments are when someone takes an idea I’ve shared, runs with it, and comes back with not only what I asked for, but something I hadn’t even considered. That’s where the magic happens.

Q: Your R&D approach starts with research-first thinking. How does that influence your company culture?

A: I’m naturally inquisitive. Every new idea starts with research—often me digging into the “why” before handing it to engineering or marketing to explore further.

We won’t build something just to compete unless we see a way to make it meaningfully better. If I can’t see that potential, I park the idea until I can. If I do see it, we’ll prototype quickly, test functionality, and only then think about aesthetics and packaging.

This approach encourages the team to think critically, test early, and be comfortable with iteration.

Q: ANTEX (Antillion’s field research activities) tests products in the wild, not just in the lab. Can you share an example where this approach paid off?

A: One example was a small device designed to extend radio range in mountainous terrain. We tested it in the Alps—hiking, throwing it in the snow, getting it damp and cold, using it the way it would actually be used.

It passed the environmental tests, but more importantly, we learned how people actually carry, deploy, and treat the equipment. That kind of insight simply can’t be replicated in a lab.

Q: As AI moves from cloud to edge, how do you see the role of field-deployed systems changing?

A: I’m still not sure how much of the current push for AI at the edge is driven by genuine need versus commercial opportunity. There are clear cases where edge AI is essential—especially when you need faster decisions or want to limit the data sent upstream—but I think the technology may be a little ahead of widespread operational readiness.

The real opportunity isn’t “AI” as a buzzword—it’s making better decisions closer to where data is created, in ways that either empower local operators or optimize system efficiency.

Q: Antillion operates in high-stakes environments where failure isn’t an option. How do you build resilience into your team?

A: We don’t really talk about “failure.” We invest heavily in early-stage research and iteration to reduce the chance of failure, but when it does happen, it’s rarely black and white. We treat it as another step toward solving the problem. Resilience comes from seeing it as a shared responsibility—between our team, our products, and our customers—to mitigate and learn from setbacks.

Q: When people look at Antillion 10 years from now, what do you hope they say?

A: I’d love for one or more of our products to still be in use, still delivering on—and maybe exceeding—their original purpose. Longevity is the ultimate proof of value. If our technology continues to serve customers in ways we couldn’t predict at launch, I’ll consider that a success.

Subscribe to our newsletter

‍

Industry leader Scott Shadley reveals how Solidigm’s innovations in SSDs, partnerships, and architecture are reshaping data centers to meet the rising demands of AI, edge, and enterprise workloads.

‍

As AI becomes more tightly integrated into applications such as robotics, manufacturing automation, and autonomous vehicles, the need for industry-specific performance benchmarks becomes increasingly important. Today, MLCommons announced it is rising to the challenge of vertically oriented benchmarking with the release of MLPerf Automotive v0.5.  

This new benchmark suite provides a trove of data for the automotive industry as its members evaluate AI systems destined for safety-critical vehicle applications. The release establishes the first standardized performance baseline for automotive AI workloads, which will help procurement decision makers in the automotive supply chain.  

Cross-Industry Collaboration Drives Innovation  

The benchmark emerged from a collaboration between MLCommons and the Autonomous Vehicle Compute Consortium (AVCC). It brings together technical expertise from organizations spanning the AI and automotive manufacturing ecosystems, including Ambarella, Arm, Bosch, C-Tuning Foundation, CeCaS, Cognata, Motional, NVIDIA, Qualcomm, Red Hat, Samsung, Siemens EDA, UC Davis, and ZF Group.  

This collaborative approach reflects the complexity of modern automotive AI systems, which must integrate everything from silicon-level optimizations to safety-critical software stacks. The benchmark addresses this reality by measuring complete system performance rather than isolated component capabilities.  

“As vehicles become increasingly intelligent through AI integration, every millisecond counts when it comes to safety,” said Kasper Mecklenburg, Automotive Working Group co-chair and principal autonomous driving solution engineer, Automotive Business, Arm. “That’s why latency and determinism are paramount for automotive systems, and why public, transparent benchmarks are crucial in providing Tier 1s and OEMs with the guidance they need to ensure AI-defined vehicles are truly up to the task.”  

Addressing Real-World Automotive Demands  

MLPerf Automotive v0.5 introduces three core performance tests: 2D object recognition and segmentation, and 3D object recognition. The tests use high-resolution, 8-megapixel imagery that reflects real-world camera systems.  

“Many of the key scenarios for AI in automotive environments relate to safety, both inside and outside of a car or truck,” said James Goel, Automotive Working Group co-chair. “AI systems can train on 2-D images to be able to detect objects in a car’s blind spot or to implement adaptive cruise control….In addition, 3-D imagery is critical for training and testing collision avoidance systems, whether assisting a human driver or as part of a fully automated vehicle.”  

The benchmark implements two distinct measurement scenarios designed for automotive contexts. The “single stream” scenario measures raw performance and throughput for applications like highway vehicle tracking. The “constant stream” scenario addresses mission-critical functions where AI systems must process data at fixed intervals, such as collision detection systems.  

Setting the Foundation for Industry Evolution  

The initial submission round included entries from NVIDIA and GATEOverflow, establishing baseline performance data for development systems (evaluation systems not inside a production vehicle) across the closed and open benchmarking divisions. The closed division enforces strict rules to enable direct apple-to-apple comparisons between systems. The open division allows more flexibility in implementation approaches, showcasing cutting-edge techniques.  

The benchmark’s impact extends beyond simple performance comparison. By standardizing measurement approaches, it promises to streamline the notoriously complex automotive procurement process, where original equipment manufacturers (OEMs) traditionally navigated a complex comparison challenge among many suppliers with limited standardization.  

The TechArena Take  

The race to implement AI in automotive just shifted into a new gear. MLPerf Automotive v0.5 creates the first neutral ground with transparent, safety-focused metrics that matter to vehicle manufacturers. Now there’s a common measuring stick with the results that can drive real procurement decisions across the global automotive market.  

For OEMs, this benchmark suite eliminates the guesswork from multi-million-dollar platform decisions. When choosing between competing AI systems for next-generation vehicles, they finally have standardized, reproducible data to base their decisions on.  

We expect the existence of these standardized benchmarks to accelerate automotive AI innovation cycles. When performance gaps become visible through standardized benchmarks, engineering teams move faster to close them. The result: better, safer AI systems reaching production vehicles sooner.

Subscribe to our newsletter.

In highly collaborative industries like media and entertainment, time isn’t just money—it’s opportunities. Giving your animators, designers, and visual effects artists more time means they have more space to coordinate and develop better creative outcomes. And when you have hundreds of collaborators, saving each one just a few minutes every hour can exponentially increase the amount of time spent on creative endeavors instead of, for example, waiting for software to load.  

I recently had the opportunity to explore how storage innovation is enhancing collaborative workflows in the media and entertainment industry with Alex Timbs, Senior Business Development Manager of Media and Entertainment at Dell Technologies, and Scott Shadley, Leadership Marketing Director at Solidigm. During our Data Insights episode, it became clear that changes in content production workflows from pre-production to final edits are causing a fundamental shift in how storage supports content creation, moving flash storage from “nice to have” to essential for modern production pipelines.

Alex brought a unique perspective to our conversation, having spent 15 and a half years at Animal Logic (now Netflix Animation) before joining Dell. His experience as the company scaled from 80 to over 1,000 people globally provided compelling real-world context for understanding storage evolution in creative environments.  

Alex saw firsthand the “serendipitous performance improvements” that emerge when organizations transition to flash storage and save minutes that add up to hours of freed-up creative time, witnessing gains that went far beyond what traditional metrics might predict. At Dell, he’s worked with customers who achieved this as well. He cited a recent Dell film studio customer who achieved 100x performance improvements—not 100% gains, but literally 100 times faster workflows.  

The need for faster storage has been recently accelerated by AI and real-time workloads, which demand rapid filling and flushing of video random access memory (VRAM) on graphics processing units (GPUs). Where 24GB VRAM used to be sufficient, today’s workloads often demand 96GB or more. To keep these GPUs fed, VRAM must be filled and flushed at extreme speeds, making high-performance flash storage no longer a luxury, but an absolute necessity.

Scott emphasized how storage has transformed from an afterthought to a critical performance enabler. The concurrent access patterns required by modern workflows—where multiple users need simultaneous access to large files alongside their associated metadata—can only be efficiently handled by flash technology. Existing spinning HDD storage simply cannot deliver the random access performance required for today’s collaborative, high-resolution content creation environments.

Dell’s AI Factory serves as a robust foundation for media and entertainment organizations striving to lead amid surging data growth, new content formats, and adoption of AI-powered workflows. The platform uniquely combines validated, full-stack solutions, enabling companies to start small and scale incrementally, directly addressing the sector’s dual mandates of technological advancement and financial discipline.

At its core, Dell AI Factory leverages the PowerScale family: from the cost-effective F210, optimized for studio or departmental use, to the high-density, high-performance F910 designed for the most demanding enterprise-scale operations. This architecture empowers customers to only pay for what they need today, with the confidence they can scale both performance and capacity linearly as their needs evolve, eliminating the risks of overprovisioning or stranded investment.

The result is a unified platform that streamlines collaborative workflows (including editing, visual effects, and broadcast), consolidates data silos, and supports both on-premises and multi-cloud deployment, all with high security and efficiency. Multiple industry-leading media organizations already rely on PowerScale for everything from 4K/8K post-production to real-time virtual production and generative AI–driven analytics. Dell’s integrated data reduction, metadata solutions, and cyber protection further drive down operational costs, while the modular “grow as you go” model enables ongoing financial prudence. This makes the AI Factory a trusted partner: future-ready, validated by top global brands, backed by deep ISV partnerships, and proven to accelerate creative delivery while protecting the bottom line.

The edge computing dimension adds another layer of complexity and opportunity. A modern film production might have 10 cameras that are capable of capturing resolutions up to 17K, and the crew will want to start working with that immediately. Alex described in-camera visual effects (ICVFX) scenarios where directors give real-time creative feedback, viewing final-quality visual effects directly on on-set monitors. This surge in edge computing for ICVFX pushes the need for high-performance storage that can operate in demanding production environments, all while delivering the rock-solid reliability that tight shooting schedules require.

Interestingly, Alex compared today’s transformation to the shift from analog film to digital photography. Just as digital cameras delivered instant feedback and removed the high cost of mistakes tied to film processing, modern workflows in content production combine real-time creative feedback with minimal risk. This immediacy allows teams to iterate more often, experiment more boldly, and ultimately achieve stronger creative outcomes by removing traditional bottlenecks.

Solidigm’s collaborative approach resonates strongly with this philosophy. Rather than pushing customers toward the highest-performance solutions regardless of need, Scott described how their solutions lab and upcoming AI lab allow customers to test workloads before making commitments. This “try-before-you-buy” model helps organizations right-size their storage investments while ensuring they can achieve their performance objectives.

Looking ahead, both experts see storage demands continuing to accelerate. Organizations working in 4K today need to prepare for native 8K workflows tomorrow, requiring storage architectures that can scale both performance and capacity over multi-year timeframes.

The TechArena Take

The convergence of AI, real-time workflows, and edge computing is fundamentally reshaping storage requirements across industries, with media and entertainment serving as the proving ground for technologies that will eventually transform other verticals. As Alex noted, the future belongs to organizations that can make the most informed real-time decisions possible, and that capability fundamentally depends on having the right storage foundation in place. Dell and Solidigm’s partnership demonstrates how thoughtful collaboration can deliver solutions that scale from individual creators to global production companies.

For more insights on Dell’s storage solutions for media and entertainment, visit their website www.delltechnologies.com/powerscale or connect with Alex Timbs on LinkedIn. Learn more about Solidigm’s AI-focused storage solutions at solidigm.com/ai or reach out via LinkedIn to Scott Shadley.

‍

The use of AI in health care promises a remarkable transformation. For an industry facing chronic staffing shortages against increasing demand, the potential for always-on support for care providers and an ability to move toward proactive, predictive care systems would literally save lives. My recent discussion with Dr. Rohith Vangalla, lead software engineer at Optum, revealed how AI has the potential to reshape everything from infrastructure architecture to clinical workflows, and why privacy-first design has become the cornerstone of scalable health care AI.

During our conversation ahead of the upcoming AI Infrastructure Summit, Rohith shared insights from his unique background, which includes backend development, aviation (he’s also a licensed helicopter pilot), and academic research. These diverse experiences have shaped his perspective that AI must focus on creating tools that make health care “smarter, faster, and more human-centric.”

The regulatory landscape in health care presents unique challenges that many industries don’t face. As Rohith emphasized, “A bad model doesn’t just mean poor performance. It literally costs lives.” Rather than viewing regulations as obstacles, he sees them as essential safety rails that prevent innovation from going off track. The real danger, he argued, lies in under-regulation that could allow biased or opaque models into clinical care, leading to misdiagnosis and eroding trust in health care systems entirely.

With trust acting as a crucial cornerstone in health care AI delivery, privacy-first architecture has emerged as an essential element to new solutions. Rohith highlighted how federated learning enables hospitals and rural clinics to train shared models without moving patient data off their servers, maintaining local data control while harnessing collective intelligence. When combined with zero-trust frameworks that verify every access request, and confidential computing that keeps data encrypted even during processing, these technologies create infrastructure that doesn’t sacrifice privacy for performance.

The conversation revealed how these architectural strategies are opening doors for international collaboration that wouldn’t have been possible otherwise. Rather than slowing down innovation, privacy-first design is actually accelerating it by enabling secure data sharing across previously isolated health care systems.

Real-world impact is already visible across multiple health care domains. AI can highlight tiny anomalies on X-rays that experienced radiologists might miss, reducing diagnostic errors and accelerating treatment. Voice-enabled documentation frees physicians to spend more time connecting with patients. And on the operational side, AI-powered call centers could route patients to appropriate specialists in seconds, eliminating anxiety-inducing hold times.

Looking ahead, Rohith identified the most exciting frontier as the shift from reactive to proactive care. Predictive analytics can now identify early risk factors for conditions like heart failure or sepsis before symptoms appear, enabling clinicians to intervene before patients require emergency care. This capability becomes even more powerful when considering underserved areas. A rural clinic without a cardiologist, for example, could leverage AI-powered tools to support general practitioners in making critical diagnoses.

The infrastructure evolution supporting these advances focuses on efficiency and accessibility rather than raw computational power. Technologies for fine-tuning large language models (LLMs) enable organizations with limited resources to customize powerful models without enormous infrastructure investments. Rohith shared an compelling example from a university hackathon where students created a lightweight AI system that could run directly in ambulances, helping paramedics triage patients en route based on vitals and symptoms without requiring cloud connectivity.

At his upcoming AI Infrastructure Summit presentation, Rohith will address the critical balance between privacy and performance when choosing cloud, on-premises, or hybrid deployments. While cloud offerings provide speed, scalability, and cost optimization, on-premises solutions offer greater control and data residency, which are crucial factors in health care. Hybrid architectures often hit the sweet spot by keeping sensitive data local while offloading heavy compute workloads to the cloud.

The TechArena Take

Rohith’s vision for health care AI represents an industry focused on practical, ethical solutions that prioritize patient outcomes. His emphasis on privacy-first architecture, responsible AI development, and proactive care models demonstrates how thoughtful engineering combined with regulatory compliance can drive meaningful innovation.

The future Rohith envisions, where AI serves as a non-judgmental health and wellness companion working in the background to ensure people feel seen, supported, and safe, reflects the true potential of AI in health care. It’s not about the technology itself, but about how that technology can bridge gaps in access, improve care quality, and ultimately save lives through early intervention and predictive insights.

Connect with Rohith on LinkedIn to continue the conversation about AI infrastructure in health care. Learn more about Optum’s AI initiatives at the Optum Marketplace, where you can find the latest articles and trials on health care AI innovation.

Surveying 250 IT pros, we found 29% already run SSDs beyond performance tiers, 81% would migrate when TCO wins, and storage innovation is a top lever to free power and space across the data center.

‍

From Intel’s layoffs to stealth automation, AI is reshaping work at a pace that outstrips human adaptation—driving record stress, uneven gains, and a scramble to reskill before the next downturn hits.

In the not-so-distant past, data center storage was somewhat of an afterthought. You needed a place to gather data; you needed it to be reliable; and you needed it to be economical. And that’s pretty much where the conversation ended. Now in the era of AI workloads, storage is taking center stage for the critical role it plays in data activation. Having the right storage solutions in the right place provides the flexibility, efficiency, and security to feed AI at scale.

I recently had the opportunity to explore this transformation with Saif Aly, senior product marketing manager at Dell, and Scott Shadley, leadership marketing director at Solidigm, to explore how enterprise storage requirements are evolving in response to AI-driven workloads and data-intensive applications. During our TechArena Data Insights episode, it became clear that storage has evolved to the critical foundation enabling AI success.

The AI workload revolution has created unprecedented demands on storage infrastructure. As Saif explained, these workloads require sustained throughput, low latency, and massive scale simultaneously. The challenge extends beyond simple performance. Enterprises face data fragmentation across edge, core, and cloud environments, creating operational complexity that can lead to vendor lock-ins and underutilized graphics processing unit (GPU) resources.

Dell’s response centers on their AI Data Platform, built on the principle that modern storage must support the entire data lifecycle. The PowerScale platform serves as the foundation, delivering what Saif described as unmatched performance improvements: 220% faster data ingestion and 99% faster data retrieval compared to previous generations. The introduction of MetadataIQ further accelerates search and querying capabilities, directly supporting AI workload requirements.

Scott emphasized how customer conversations have evolved beyond traditional capacity discussions to focus on “time to first data”—how quickly organizations can access information when they need it. In AI application workloads, different data types require varying levels of accessibility and performance characteristics. The challenge lies in understanding what data needs to sit directly adjacent to GPUs versus what can be retrieved from more distant storage tiers.

The discussion revealed how inference workloads, particularly retrieval-augmented generation (RAG) architectures, create unique storage demands. These systems require large datasets to be readily accessible for real-time referencing while simultaneously managing active data processing next to compute resources. Success depends on optimizing the balance between high-performance local storage and efficient data movement from archive locations.

While flash storage dominates high-performance applications, both experts acknowledged that hard disk drives (HDDs) retain value for cold and warm datasets. The key insight: not all data is equal, and successful architectures blend flash-based solid-state drives (SSDs) and HDD storage within unified namespaces to balance performance and cost considerations.

The conversation highlighted remarkable capacity evolution, with Saif recounting his amazement at holding Solidigm’s 122 TB drive, a device containing massive data volumes in a small form factor. This density revolution, progressing from 30 TB to 60 TB to 122 TB drives just in the last year, enables dramatic improvements in rack space efficiency, power consumption, and cooling costs while maintaining the throughput AI workloads demand.

Scott connected this capacity evolution to practical customer needs, explaining how optimization now focuses on the right bandwidth, density, and time-to-data characteristics rather than simply maximum speed. As storage capacity per device increases, the focus shifts to infrastructure optimization that delivers customer value through improved total cost of ownership and operational efficiency.

Real-world impact emerged through customer examples Saif shared. Kennedy Miller Mitchell, the studio behind the Mad Max franchise, used PowerScale to enable pre-visualization of entire scenes before filming. That capability allows directors to iterate creatively and make real-time decisions. Subaru leveraged the platform to manage exponentially growing data volumes, handling 1,000 times more files than previously possible and directly improving their AI-driven driver-assistance technology accuracy.

Looking ahead, both experts see storage demands continuing to accelerate, driven by AI’s exponential data growth and evolving workload requirements. As Saif noted, “the data explosion is not going to stop,” with AI both consuming and creating massive amounts of data. The distributed nature of modern computing—spanning edge, core, and cloud environments—requires storage solutions that provide consistent experiences and seamless data mobility across all locations.

The TechArena Take

The convergence of AI workloads, massive data growth, and distributed computing architectures is fundamentally reshaping enterprise storage from a cost center to a strategic enabler. Dell and Solidigm’s partnership demonstrates how thoughtful collaboration can deliver solutions that scale from individual creators to global enterprises while addressing the critical balance between performance, capacity, and cost efficiency. As storage continues to assert its place as a foundation of modern workloads, organizations that invest in flexible, high-performance architectures today will be best positioned to capitalize on tomorrow’s AI-driven opportunities.

For more insights on Dell’s enterprise storage solutions, visit Dell.com/PowerScale or connect with Saif Aly on LinkedIn. Learn more about Solidigm’s AI-focused storage innovations at solidigm.com/AI or reach out via LinkedIn to Scott Shadley.

Watch the podcast

Subscribe to our newsletter

‍

Allyson Klein and Robert Blum of Lightwave Logic unpack how electro-optic polymers, paired with silicon photonics, lower power and boost density on the road to AI-fueled 400G-per-lane optics—with a 2027 volume ramp in sight.

TechArena’s flagship interview series earned a 2025 Stevie® Award in the International Business Awards® annual contest, recognizing authentic, executive-level conversations on AI, data centers, edge, and sustainability.

The In the Arena podcast took home recognition in the highly competitive “Shows – Technology” category. The win acknowledges a simple idea that has guided the series from day one: put real innovators in the spotlight and make complex technology understandable for decision-makers. It’s a gratifying milestone for a show that has quietly grown into a go-to forum for leaders shaping the infrastructure of the AI era.

Now in its 22nd year, the IBAs are widely regarded as one of the world’s premier business awards programs, drawing thousands of entries from organizations of all sizes and across industries. Winners are selected by the average scores of more than 250 global executives, with this year’s judging taking place from May through July.

In the Arena is hosted by industry veteran Allyson Klein—TechArena’s founder and principal—and produced by the TechArena editorial team. Across three seasons, the show has welcomed executives and founders from companies representing more than $9 trillion in market capitalization—including Microsoft, NVIDIA, Google, and Meta—alongside standout startups pushing the edges of compute, storage, networking, and energy. The conversations are deliberately accessible without sacrificing depth, aiming to surface the real decisions behind AI deployments, data center architectures, and sustainability initiatives.

Judges called out the podcast’s high production quality, editorial clarity, and guest caliber, noting that the series “bridges executive insight with emerging innovation.” Several praised the show’s consistent focus on AI and sustainability, and highlighted its effort to elevate underrepresented voices in tech leadership—an intentional part of the booking strategy since the series’ inception.

For TechArena, the recognition is as much about community as it is about content.

“This recognition celebrates the innovators who’ve delivered insights across our 176 episodes and our audience, which values the conversation as technology accelerates in the era of AI,” Klein said.

Beyond the accolades, the show’s impact shows up in the themes it consistently returns to: the operational realities of AI at scale; the fast-changing power and cooling profiles behind GPU clusters; the role of high-capacity flash and data orchestration in breaking bottlenecks; the promise (and limits) of optical I/O; the emergence of agentic workflows in semiconductor design; and the practical steps organizations can take to reduce environmental impact while improving performance. The result is a library designed for CTOs, architects, and product leaders who need both strategic direction and applied lessons, not just highlight reels.

The win also comes at a moment when enterprise leaders are hungry for grounded dialogue. AI adoption has accelerated, but success still hinges on fundamentals: data placement, latency, reliability, energy, and cost. In the Arena aims to be a steady companion through that transition—candid, technical when it matters, and always human.

Looking ahead, the team is lining up episodes leading into upcoming conferences including Yotta, AI Infra, OCP Global and SC’25.  

To everyone who’s listened, shared an episode, or joined us behind the mic: thank you. And to the judges—thank you for recognizing a show built on curiosity, clarity, and respect for the people doing the work.

From federated learning and zero-trust to confidential computing, Dr. Rohith Vangalla shares a practitioner’s playbook for explainable, scalable AI that moves healthcare from reactive to proactive.

For every headline celebrating agentic AI’s potential to revolutionize business, there’s a data privacy lawsuit quietly working its way through the courts—a reminder that innovation has outpaced consent infrastructure. This week, Permission announced Permission Agent, a system designed to broker high-quality human data with verifiable consent and contributor rewards.  

“AI is only as good as the data it’s trained on, and the best data comes directly from people—with their permission,” said Charlie Silver, CEO of Permission. “Permission Agent is the missing bridge between individuals and AI systems, enabling direct, compliant, and mutually beneficial data exchange at scale.”

How Permission Agent Works

Permission Agent operates as a persistent, identity-tied “digital mini-me.” It collects only user-approved signals (e.g., intent, preferences, context) and attaches usage rights and consent metadata to each record. Buyers receive structured datasets with provenance and audit trails so they can prove lawful basis and honor revocation. Contributors are compensated in $ASK, which Permission has made omnichain via LayerZero’s OFT standard, allowing movement across supported chains without wrapped tokens.

Why This Matters Now

Enterprises racing into agentic architectures are discovering their identity and governance foundations weren’t built for autonomous actors. Without machine-readable consent and revocation, agents can overstep policies and contracts, raising legal exposure as author and media cases proceed and as regulations (from GDPR to the EU AI Act) tighten expectations for transparency and consent.

Market Context

Permission isn’t the first to reward data contributors, but it’s one of the few aiming squarely at consented human signals for AI rather than a single vertical. Where Brave compensates attention and projects like Hivemapper/DIMO target maps and vehicle telematics, Permission’s pitch is a portable, auditable consent layer and marketplace that personalization teams and AI builders can safely use.

Permission Agent is in early access for enterprises and individual contributors. AI organizations can request sample datasets; consumers can join the waitlist.

A Buyer’s Checklist for Evaluating Consent-First AI Data Platforms

For enterprises considering purchasing permissioned data for AI, here are some due diligence suggestions:

• Start with provenance and consent semantics. Require a documented consent-scope model tied to specific data elements and enforceable purpose limitation, plus a true right to withdraw at the row, feature, and embedding level—complete with deletion that propagates to vector indexes and models.
• Treat agent identity and guardrails as a first-class security domain: autonomous agents should have unique credentials, least-privilege, time-bounded scopes, step-up authentication for sensitive actions, and human-in-the-loop approvals.
• For audit and forensics, require an immutable chain of custody that captures when and where each datum was collected, the consent state at capture and at use, and every downstream touchpoint, with machine-readable deletion receipts confirming propagation to feature stores, embedding indexes, caches, and backups.
• Scrutinize reward mechanics and compliance: how contributor rewards are calculated and distributed across chains; how custody and transfers are secured; and how KYC/AML, tax reporting (e.g., 1099-DA in the U.S.), consumer-protection, dispute-resolution, and contributor communications are handled.

The TechArena Take

Agentic AI doesn’t scale without machine-consumable consent. Permission’s move is notable for putting the individual at the center of a permissioned data supply chain aimed at training and personalization—and for bundling compensation and auditability from the start.

Two execution risks will determine impact:

Adoption density: Quality requires sustained, diverse contributors and buyer demand—plus SDKs that make capture, revocation, and downstream enforcement trivial across RAG, fine-tuning, and personalization stacks.

Enforceable revocation: Recording consent isn’t enough. Buyers will expect deletions to propagate to feature stores, embeddings, caches, and logs—fast.

Bottom line: If Permission proves dataset quality, scalable revocation, and low-friction integration, it could become the default vendor for “human signals with receipts” in agentic pipelines—good for builders, and overdue for the people whose data fuels the system.

Want to learn more? Check out www.permission.ai/.

Verizon Business’ Nikhil Tyagi shares insights on scaling AI at the edge—from small language models and multimodal experiences to infrastructure challenges and adaptive inference. Want to learn more about the AI Infra Summit and see Nikhil Tyagi live? Find out more here.

The data center industry faces an ongoing challenge: how to securely reuse storage devices when decommissioning them without compromising data integrity. My recent Great Debate with Amber Huffman and Jeff Andersen from Google revealed not just the scope of this challenge, but how the Open Compute Project’s Layered Open-Source Cryptographic Key-management (OCP L.O.C.K.) initiative could reshape how the industry approaches storage security and sustainability.

During our discussion, Amber and Jeff painted a picture of an industry with a dilemma. Everyone would like to get longer lives out of storage devices. But to date, there hasn’t been a solution that still meets the top priority of protecting users’ data. As Amber explained, in the second-hand market, even encrypted drives face potential threats from nation-state actors with substantial resources, and evolving technology could eventually break older encryption algorithms.
In current practices, organizations physically destroy drives or perform time-consuming multi-pass overwrites. Destruction, while secure, creates significant operational inefficiencies and environmental waste; overwrites take a long time and are failure prone.

OCP L.O.C.K., currently at 0.85 specification and available for review, is a new alternative, a comprehensive project to deliver an open implementation at CHIPS Alliance that provides encryption key management services to storage drives and hosts. It builds on the established Caliptra open-source hardware root of trust, also an implementation at CHIPS Alliance, and will be integrated into Caliptra 2.1.

OCP L.O.C.K. improves on traditional data security methods in several ways. At its core, OCP L.O.C.K. ensures that only trusted, verified components can access the encryption keys that protect data on drives. The system creates multiple layers of key management. So when a drive is provisioned with OCP L.O.C.K., cloud service providers can trust that data remains inaccessible without the proper access credentials. And when that drive needs to be decommissioned, OCP L.O.C.K. attests that process has been completed successfully.

Jeff’s and Amber’s insights into the technical architecture revealed the sophistication of this approach. OCP L.O.C.K. introduces multi-party authorization, requiring multiple keys rather than a single password to access drive data. This creates layers of protection that persist even if a drive is physically stolen. The implementation leverages Caliptra’s open-source foundation, allowing the security community to harden these systems through collaborative development.

The possibilities created by OCP L.O.C.K. have broad implications for the storage ecosystem. Amber emphasized how OCP L.O.C.K. could transform the value chain, enabling hyperscalers and businesses to sell decommissioned drives in secondary markets rather than destroying them. This represents a significant shift toward sustainability without compromising security — achieving what she called “the best of both worlds.”

The technical roadmap they outlined demonstrates the project’s maturity and industry backing. With partners including Microsoft, Samsung, Kioxia, and Solidigm, along with contributions from other industry partners, OCP L.O.C.K. aims to be a standard implementation rather than a niche solution. Work with standards bodies like Trusted Computing Group (TCG) has already yielded results, with the OCP L.O.C.K.-inspired Multiparty Authorization specification having now been published. These efforts, along with targeting Federal Information Processing Standards (FIPS) compliance, show the thorough approach being taken to ensure widespread adoption.

Looking toward the future, both speakers highlighted how OCP L.O.C.K. represents a broader trend in hardware security. Amber’s observation about the shift from open-source software to open-source hardware particularly resonated, suggesting we’re entering an era where foundational security components that aren’t differentiators become collaborative, community-driven efforts rather than proprietary implementations.

The implications for post-quantum security were equally compelling. Jeff’s discussion of hybrid cryptographic approaches, combining established elliptic curve methods with new post-quantum algorithms like ML-KEM (module-lattice-based key-encapsulation mechanism), demonstrates how OCP L.O.C.K. is designed not just for today’s threats but for the quantum computing era ahead.

What’s the TechArena take? The OCP L.O.C.K. initiative represents more than just another security standard: it’s a reimagining of how the industry approaches storage lifecycle management. By combining robust security with environmental responsibility, Google and its partners are creating a framework that could serve as a model for other infrastructure components. As data centers continue to scale and sustainability becomes increasingly critical, initiatives like OCP L.O.C.K. show how collaborative open-source development can address both security and environmental challenges simultaneously.
The project’s emphasis on implementation rather than just specification, combined with its open-source approach and broad backing, suggests we’re looking at a future where secure storage decommissioning becomes as routine and trustworthy as Amber hopes — ultimately “boring” in the best possible way.

Connect with Amber Huffman and Jeff Andersen on LinkedIn to follow their continued work on storage security and open compute initiatives. The OCP L.O.C.K. 0.85 specification is available for download today, with the 1.0 release targeted for later this year ahead of the OCP October global summit.

Listen to the podcast

‍

The Ai4 conference kicked off today at the MGM Grand in Las Vegas, drawing an expected 8,000+ attendees for three days of keynotes, 600+ speakers across 50 tracks, and a 250-plus-vendor expo aimed squarely at applied, enterprise AI. Organizers touted a sharpened focus on agentic systems, governance, and real-world deployments.

Opening day set the tone with two back-to-back arena sessions: a fireside chat on AI’s impact on classrooms and childhood with AFT president Randi Weingarten, followed by Tengyu Ma, Chief AI Scientist at MongoDB, on “RAG in 2025: State of the Art and the Road Forward.”

What’s new this year

Ai4’s program stretches across 19 stages and 50 themed tracks, including Generative AI, Agentic Systems, AI Policy, a reimagined AI Research Summit, and a “Beginner’s Summit” to onboard first-timers. Exhibit-hall action will formally begin Tuesday morning and run alongside daily receptions.

Tuesday’s opening main stage will feature Shirin Ghaffery of Bloomberg News moderating a fireside chat with Geoffrey Hinton, who is known as the visionary godfather of AI. The talk is titled "AI, Ethics, and the Future of Humanity." The event will highlight a slate of sector tracks, live demos across the show floor, and a social-impact spotlight with Colossal Biosciences’ Ben Lamm discussing de-extinction and AI in a fireside format.

Wednesday’s keynotes will lean into the future of perception and agents. Fei-Fei Li will dive into world models, spatial intelligence, and human-centered AI, followed by a Cisco fireside where President & CPO Jeetu Patel will explore what it will take to safely catalyze an “agentic AI revolution.”  

For buyers: where the rubber meets the road

If you’re attending Ai4 to evaluate products, the expo floor is positioned as more than a showcase. Expect hands-on demos spanning AI applications & agents, data platforms, security/governance, and cloud infrastructure—plus brand activations.

Signals to watch

With launches and briefings from governance vendors and platform players, expect concrete approaches to policy, observability, and controls for autonomous workflows.

Ma’s keynote framed how retrieval augments, rather than competes with, fine-tuning and long-context—important for enterprises balancing cost, latency, and source-of-truth requirements.

With Fei-Fei Li headlining, spatial intelligence and world modeling will continue to pick up steam as the path beyond text-only systems—especially for robotics and simulation-heavy industries.

The TechArena Take

Ai4’s program reads like a referendum on agentic AI at enterprise scale. Three practical threads stand out for IT and cloud architects:

• Control planes for agents will define winners. The next six months won’t be about flashiest copilots—it’ll be about who ships the cleanest hooks for policy, identity, data-scope, and human-in-the-loop stops across multi-agent workflows. If you can’t prove guardrails, you won’t get production sign-off.
• Data gravity beats model gravity. Ma’s framing of RAG vs. finetune vs. long-context mirrors what we’re hearing from large shops: push reasoning to the edge of your governed data layer; keep PII and contracts in-bounds; and reserve heavyweight finetunes for workloads that truly demand it.
• Infra is the quiet story. As agents expand from chat to autonomous task execution, winning stacks will hinge on event-driven integration, vector/search performance, observability, and low-latency networking—plus storage patterns that keep embeddings and history auditable.

The enterprise AI landscape is undergoing a fundamental transformation. While organizations have focused heavily on graphics processing unit (GPU) compute power and model sophistication, a critical infrastructure component has emerged as the new performance differentiator: storage. The Supermicro Open Storage Summit, running from August 12 to 28 with online sessions from leading solutions providers, promises to reveal how innovative storage strategies are delivering breakthrough performance improvements that could reshape your AI deployment economics.

The Hidden Performance Multiplier in AI Infrastructure

As organizations scale from AI experimentation to production deployment, they’re discovering that inference workloads demand different storage characteristics than training pipelines. The data tells a compelling story: enterprises deploying solid state drive (SSD) storage solutions are seeing 10x to 20x throughput improvements, 4,000x input-output per second (IOPS) scaling improvements, and up to 40% total cost of ownership (TCO) reductions compared with traditional storage solutions.

These aren’t theoretical gains. Real-world implementations for retrieval-augmented generation (RAG) workloads have demonstrated that storage optimization with SSDs can deliver 70% increases in queries per second while simultaneously reducing memory footprint by 50%. For enterprises struggling with the economics of AI deployment, these performance multipliers represent an opportunity to maximize return on investment.

Two Critical Sessions You Can’t Afford to Miss

The Supermicro Open Storage Summit expands on these opportunities with two must-attend sessions that tackle the most pressing storage considerations facing enterprise AI deployments today.

Storage to Enable Inference at Scale (August 19, 10:00 AM PT) brings together industry leaders from Solidigm, Supermicro, NVIDIA, Cloudian, and Hammerspace to explore how new storage protocols and distributed inference frameworks are enabling large-scale inference processing. This session will reveal how organizations are moving beyond traditional storage approaches to deploy validated infrastructure optimized for GPUs that unlocks real-time performance at scale.

Enterprise AI Using RAG (August 27, 10:00 AM PT) dives deep into RAG, one of the most critical enterprise AI use cases. With experts from Solidigm, Supermicro, NVIDIA, VAST Data, Graid Technology, and Voltage Park, this session addresses how enterprises can operationalize generative AI securely and efficiently while maintaining proximity to their most valuable data assets.

The SSD Revolution: Beyond Traditional Storage Thinking

One of the most compelling insights emerging from enterprise AI deployments challenges conventional storage wisdom. Solidigm’s recent breakthrough work, which will be discussed in the upcoming sessions, demonstrates that strategically offloading data from memory to high-performance SSDs doesn’t just reduce costs: it actually improves performance in many scenarios.

The company’s innovative approach involves moving model weights and RAG database components from expensive distributed random-access memory (DRAM) to optimized SSDs, achieving better performance at lower cost. In one demonstration involving a 100 million vector dataset, this approach delivered 57% less DRAM usage while maintaining or even improving query performance. The economic implications are huge as enterprises can run complex models on GPUs that would otherwise lack sufficient onboard memory.

From Data Center Footprint to Power Efficiency: The Complete TCO Story

The storage optimization story extends far beyond raw performance metrics. In the upcoming sessions, Solidigm will also discuss how cutting-edge storage solutions are demonstrating dramatic improvements in TCO across the entire infrastructure stack.

Take a practical example, a 50-petabyte dataset deployment with 12 NVIDIA H100 systems. Traditional HDD-based approaches require nine racks consuming 54 kilowatts. Deploy high-density 122TB SSDs, and that footprint shrinks to a single rack with up to 90% power reduction and 50% increase in available GPU footprint.

These efficiency gains matter more than ever as enterprises grapple with data center space constraints, cooling challenges, and escalating power costs.

The TechArena Take: The Storage Revolution Is Here

Organizations that leverage cutting-edge storage optimization strategies are positioning themselves for sustainable competitive advantage. While competitors struggle with infrastructure costs and performance limitations, early adopters are achieving superior AI outcomes at lower total cost of ownership.

The ability to deploy more sophisticated models, process larger datasets, and deliver faster inference responses directly translates to better customer experiences and operational efficiency.

The window for competitive advantage is narrowing rapidly. As these storage optimization techniques become mainstream, the organizations that implement them first will establish performance and cost advantages that become increasingly difficult for competitors to match.

Register Now

The Supermicro Open Storage Summit provides an opportunity to learn directly from teams of industry leaders who are defining the future of AI infrastructure. With sessions featuring experts representing all layers of the stack, you’ll gain access to the collective expertise of the companies driving AI infrastructure innovation. The summit’s focus on real-world implementations, demonstrated performance improvements, and practical deployment strategies makes it essential viewing for any organization serious about scaling AI effectively.

Don’t let storage bottlenecks limit your AI ambitions. Register below today and discover how strategic storage optimization can transform your enterprise AI performance while dramatically improving your deployment economics.

Storage to Enable Inference at Scale | August 19, 10:00 AM PT

Enterprise AI Using RAG | August 27, 10:00 AM PT

‍

I still remember walking through the bustling electronics markets of Tokyo and Hong Kong with Jim Pappas, marveling at the incredible diversity of devices and form factors surrounding us. As my manager early in my Intel career, Jim had a unique way of opening my eyes to the bigger picture—showing how the foundational work we were doing in semiconductor and standards development was sparking creativity and innovation across the globe.

This week, as the Future of Memory and Storage (FMS) conference concluded in Santa Clara, I was thrilled to see Jim receive the event’s most prestigious recognition—the 2025 Lifetime Achievement Award. As someone who witnessed firsthand his passion for the vibrant innovation that happens through standards delivery of technology, watching the industry honor his decades of foundational contributions felt like a full-circle moment.

Jim has had a tremendous impact on the industry throughout the trajectory of his career. Just imagine a world without a USB standard for a second...or without PCI and all its variants. These technologies that Jim and his collaborators helped define have formed a foundation for compute innovation. I was recently at a Tech Field Day event where I listened to the delegates sharing their Jim Pappas stories, how each of them had been impacted by his leadership over the years. One thing I was lucky enough to know is that beyond his technical brilliance, Jim’s approach to management was transformational. For me personally, he enabled me to take risks, provided incredible autonomy, and helped accelerate my career and set me up for success. To FMS: Well done on this recognition. To Jim: thanks for all that you did to propel a young Allyson forward. Your impact informs my work every day.

AI Takes Center Stage at FMS 2025

The conference that honored Jim’s legacy also marked a pivotal moment for the memory and storage industry, as AI dominated the conversations in Santa Clara. In fact, FMS 2025 featured AI in more than 60% of its keynote presentations and expert panel sessions—a clear signal that AI workloads are fundamentally reshaping the entire storage industry.

“Artificial intelligence is no longer just part of the conversation—it is the conversation,” said Tom Coughlin, Conference Chair of FMS. The three-day event brought together industry leaders to explore how innovations in DRAM, NAND, CXL, and computational storage are revolutionizing AI inference and training at scale.

Honoring Three Decades of Standards Leadership

Jim Pappas’s recognition with the Lifetime Achievement Award represents more than individual accomplishment—it celebrates the foundational standards work that enables the entire industry to thrive. His journey began in 1991 with establishing the PCI-SIG and PCI standard, work that later evolved into defining the PCI Express (PCI-e) standard that remains foundational to computing platforms today.

After joining Intel in 1994, Jim collaborated with seven companies—including IBM and Microsoft—to create the hugely successful USB standard. His recent contributions include launching SNIA’s Persistent Memory Technology Initiative, chairing the Compute Express Link (CXL) Consortium, driving formation of Universal Chiplet Interface Express (UCIe), and serving as President of Ultra Accelerator Link (UAL) for AI scale-up architectures.

“For over 30 years, Jim Pappas has played a pivotal role in creating a number of the most important standards and industry organizations which have been critical in the dramatic growth of the memory and storage industries,” Coughlin noted during the award presentation.

Having worked directly with Jim, I can attest that his influence extends far beyond technical standards. He understands that true innovation happens through collaboration and ecosystem building—principles that shaped not just technologies like USB and PCI, but entire generations of technologists who learned from his example.

Record-Breaking Innovation Recognition

FMS 2025 concluded with the presentation of its Best of Show Awards, with the program receiving a record-breaking number of nominations in its 19th year. The awards showcased the breadth of innovation across memory and storage technologies, particularly highlighting solutions addressing AI infrastructure challenges.

Celebrating Women in Technology

AMD’s Rita Gupta earned the 2025 SuperWomen of FMS Award, sponsored by Hammerspace and Pure Storage. As a Fellow in AMD’s Server System Architecture team and CXL End-End Architect, Gupta leads advanced CXL memory system architectures for current and next-generation EPYC platforms including Genoa, Turin, and Venice.

Her impact extends beyond AMD as co-chair of the CXL Consortium Memory Systems Workgroup, where she’s helped shape industry standards by authoring JEDEC CMC01 and JESD325 specifications and contributing significantly to CXL 2.0 and 3.0 development. Gupta’s work exemplifies the collaborative, standards-driven approach that Jim Pappas championed throughout his career.

Innovation Across the Storage Spectrum

The Best of Show Awards highlighted breakthrough innovations across multiple categories:

• Pure Storage earned recognition with DirectFlash QLC winning the Most Innovative Hyperscaler Implementation award, showcasing leadership in enterprise flash storage optimized for large-scale deployments. The technology represents a significant advancement in QLC NAND deployment for enterprise workloads requiring massive capacity and cost efficiency.
• Solidigm won the Most Innovative Technology award for its Liquid Cooled Hot Swappable NVMe SSD in the SSD Technology category, addressing critical thermal management challenges in high-performance AI and HPC environments. This innovation enables continuous operation at peak performance levels while maintaining system reliability through advanced cooling integration.
• VAST Data secured the Most Innovative Artificial Intelligence (AI) Application award for its collaboration with the NHL® in the Media and Entertainment Solution category, highlighting the intersection of AI and entertainment technology. The platform demonstrates how AI-driven storage solutions can transform real-time sports analytics and fan engagement experiences.
• Verge.IO claimed the Most Innovative Startup Company award for VergeIQ in the Virtualization IT Infrastructure category, demonstrating the continued evolution of virtualized infrastructure solutions. Their platform showcases how next-generation virtualization can simplify data center operations while reducing infrastructure complexity.

Additional notable winners included Micron for 1-Gamma Node LPDDR5X LPDRAM, Samsung for PM1763 16-Channel PCIe Gen6 SSD, KIOXIA for LC9 Series 245.76 TB SSD with BiCS FLASH generation 8 Memory, and Western Digital for Advanced Rare Earth Material Capture Program.

Standards as Innovation Enablers

A significant theme throughout FMS 2025 was the critical role of industry standards in enabling the AI revolution. Multiple awards recognized standards organizations, including the CXL Consortium for CXL 3.X Specifications, UALink Consortium for UALink 200G 1.0 Specification, and various SNIA technical work groups.

These standards ensure interoperability and scalability as the industry races to meet exponentially growing AI workload demands. The emphasis on standardization reflects industry maturation and recognition that collaborative approaches—the kind Jim pioneered with USB and PCI—remain essential for addressing complex technical challenges.

The TechArena Take

As FMS 2025 concluded, it became clear that the memory and storage industry has firmly embraced its role as a foundation for the AI revolution. The convergence of advanced memory technologies, innovative storage architectures, and industry-wide standardization efforts positions the sector for continued rapid growth as delivery of data across an AI pipeline becomes pervasively critical to organizations.

The record-breaking attendance and award nominations demonstrate the vitality and innovation driving the industry forward. With AI workloads continuing to evolve and scale, the memory and storage ecosystem will remain at the forefront of enabling next-generation computing capabilities.

Reflecting on Jim Pappas’ recognition and the broader innovations showcased at FMS 2025, I’m reminded of those walks through Tokyo electronics markets years ago. Jim’s vision of how foundational standards could spark global creativity and innovation has proven remarkably prescient. Today’s AI revolution builds directly on the collaborative, standards-driven approaches he championed—USB ports powering development workstations, PCI Express connecting GPUs and accelerators, and newer standards like CXL enabling the memory architectures that make modern AI possible. With so much grappling on how standards keep pace for AI innovation, I think its prescient to remember how we all benefit from this collaborative innovation and not be swayed by a need for stovepiped custom designs for bespoke deployments.

My final takeaway is how memory and storage are climbing into the center of the AI conversation. Without innovation in this space, costly GPUs can spend time in idle waiting for data delivery and costing organizations wasted cycles and opportunity. FMS is part of this solution. As Coughlin noted, “FMS is the place where the entire ecosystem meets to solve these challenges head-on.” The 2025 event proved that this industry, built on foundations laid by pioneers like Jim Pappas, continues to rise and meet each new technological moment with collaborative innovation and unflinching determination.

Dell and Solidigm leaders explore how modern storage—flash, SSDs, and flexible architectures—enables AI, accelerates performance, and helps enterprises manage data across edge to cloud.