5YF Episode #31: Voltron Data CEO Craig Dunham

Future Of Data: Unlocking The Hidden 90%

subscribe and listen today

Today we dive into the future of turning data into value, at scale.

Data infrastructure isn’t just about hardware—it’s about the intelligence built on top of it. Over the next five years, we will see a radical transformation in how data is processed, stored, and utilized, driven by AI, compute efficiency, and new architectures. But while the conversation often centers on hardware—GPUs, data centers, and even the $500 billion Stargate project—huge breakthroughs are happening already in the software and systems that sit on top of that infrastructure.

It’s awakening companies to the art of the possible.

In my conversation with Craig Dunham, CEO of Voltron Data, we explored the hidden inefficiencies in today’s data systems and what companies are doing to unlock new levels of performance. From untapped petabytes of enterprise data to the rise of energy-efficient compute, we discuss the future of data infrastructure and how it will redefine business, AI, and the global economy.

My 5 Year Outlook:

• Gains Delivered By The Invisible Processing Layer: Step change in compute output and data processing thanks to advances in the ‘invisible’ processing layer.‍
• AI Models Will Be Hungry For Smarter Compute: Software optimizations down to the bare metal of GPUs and data centers to unlock smarter compute, not just more compute.‍
• Agentic Insights Opens Door To Fully Autonomous Decision-Making: The world will move from managing AI to trusting AI.

Curious? Read on as I unpack each below 👇🏼

Gains Delivered By The Invisible Processing Layer

90% of the data that organizations have just goes unused.

For decades, companies have built massive compute infrastructures, yet 90% of enterprise data remains untapped. The problem isn’t storage—it’s processing bottlenecks that make using this data too slow, too expensive, and too fragmented.

The next five years will bring the rise of “invisible processing”—highly optimized, automated data pipelines that seamlessly process and serve insights in real time, without human intervention. Voltron Data is already proving this with claims that its customers can reduce their server usage from 200 to 2, cutting costs by orders of magnitude. Theseus, its GPU-accelerated query engine, is an early glimpse of this shift, proving that data processing can be done 100x faster and at a fraction of the cost.

This shift will help alleviate traditional infrastructure concerns and allow companies to focus entirely on real-time decision-making rather than managing data pipelines. Instead of moving vast amounts of data across different cloud environments, processing will happen where the data lives, reducing cost and complexity. AI models will no longer be trained solely on historical data; instead, they will continuously ingest and learn from live enterprise data streams, unlocking a new level of intelligence across industries.

For example banks that struggle with fraud detection at scale need to process billions of transactions per day, but traditional data pipelines couldn’t keep up. By optimizing data processing at the GPU level, they are able to run models in real time rather than waiting hours or days for batch processing. This is the future of compute—instant, automated, and embedded into every transaction.

Craig Dunham, CEO of Voltron Data

Voltron Data is a company at the forefront of accelerating data processing for AI, analytics, and enterprise-scale workloads. Voltron provides the infrastructure necessary to handle enormous amounts of data — transforming bottlenecks into breakthroughs. By championing open-source frameworks like Apache Arrow, Voltron is building the connective tissue that allows businesses to process data at orders-of-magnitude speed and efficiency, reshaping industries from finance to healthcare to national security — partnering with the likes of Snowflake and Meta. Voltron have established themselves as a key part of the AI infrastructure stack and have raised a total of $110M from the likes of Coatue, LightSpeed, Google Ventures and BlackRock.

Craig Dunham is Voltron’s CEO and brings a deep background in scaling data infrastructure businesses. Before Voltron, Craig was the CEO of Lumar, a leading SaaS technical SEO platform. Prior to that, he held significant roles including General Manager at Guild Education and Seismic, where he led the integration of Seismic’s acquisition of The Savo Group and drove go-to-market strategies in the financial services sector. Craig began his career in investment banking with Citi and Lehman Brothers before transitioning into technology. He holds an MBA from Northwestern University’s Kellogg School of Management

AI Models Will Be Hungry For Smarter Compute

The AI boom has driven unprecedented demand for compute, pushing data centers to their limits in power, space, and cost. Jensen Huang of NVIDIA has warned that AI inference could require 10x more compute power than training—a trajectory that, if unchecked, would make today’s AI infrastructure completely unsustainable.

But the future of AI isn’t just about more GPUs—it’s about smarter compute. Instead of brute-force scaling, companies will focus on optimizing existing hardware for higher efficiency and output. DeepSeek’s GPU optimizations have shown what’s possible when software fully exploits hardware capabilities, and Groq’s LPUs are further proving that rethinking chip architecture can unlock orders of magnitude performance gains.

All eyes will be on Elon Musk’s Memphis data center and the next iteration of Grok as test cases for how tightly integrated hardware-software stacks can redefine AI compute. This signals a shift toward engineers working closer to the metal, where fine-tuning infrastructure at the lowest levels will become a competitive advantage.

In the next five years, we’ll see a new wave of AI-first databases, query engines, and processing frameworks that intelligently allocate compute based on workload needs, rather than the industry’s current “buy more GPUs” mentality. Governments and cloud providers may shift from building the biggest AI supercomputers to the most efficient ones, sparking a new arms race in compute optimization.

Craig Dunham highlighted a key reality: most enterprises are sitting on massive amounts of unused compute capacity. Their existing servers aren’t fully utilized, yet they keep buying more infrastructure. The companies that win the next phase of AI won’t be the ones with the most hardware—they’ll be the ones that unlock the full potential of what they already have.

‍

Agentic Insights Opens Door To Fully Autonomous Decision-Making

As these agents become more widespread, the demand for data and analytics will surge alongside of it. They'll request and process information from large scale data sources at a rate far exceeding that of human analysts.

The rise of AI agents is redefining how businesses operate, accelerating the demand for higher-speed data processing at lower economic cost. These agents aren’t just assisting humans; they are beginning to take over real-time decision-making, pulling insights from massive data sets and acting instantly.

Today, predictive analytics helps companies anticipate trends, but humans still make the final call. That’s about to change. By 2030, enterprises won’t just use AI to predict the future—they’ll use it to act on it automatically. AI-driven systems will optimize supply chains, financial trading, cybersecurity, and marketing without human oversight.

As businesses move from analyzing past data to continuously optimizing the present, AI will execute high-stakes decisions at a scale humans simply cannot match. Systems will autonomously adjust pricing, manage inventory, detect fraud, and trigger security responses in milliseconds. This shift will demand a new level of compute efficiency, as AI agents will require instant access to vast amounts of structured and unstructured data to operate effectively. Governments and regulators will be forced to rewrite the rules around AI-driven decision-making in finance, healthcare, and national security.

Consider a government agency using AI-powered geospatial analytics. Instead of waiting weeks for human analysts to review satellite imagery, AI will autonomously detect anomalies in real-time—flagging potential security threats, disasters, or climate risks. But it won’t stop there. AI agents will trigger automated workflows, such as deploying drones, adjusting emergency response plans, or rerouting resources before a crisis escalates.

To make this future a reality, AI agents must be powered by dramatically faster and more efficient compute. Running these systems at scale will require an order-of-magnitude reduction in data processing costs—something that cannot be achieved with today’s brute-force hardware approach. The future of AI decision-making hinges on smarter compute, optimized infrastructure, and real-time data pipelines.

This is the next frontier of data processing—where AI shifts from analyzing the past to actively shaping the future. With data volumes exploding and AI models becoming more sophisticated, the benefits of giving up manual control to autonomous systems will soon outweigh the risks. The world will move from managing AI to trusting AI.

Time to get crunching on that data!

‍