In parallel to the exponential rise in demand for data and analytics, the legacy giants have lost their once-unquestioned edge in query performance—leaving an opening for competitors to dominate not only on ease of use and cloud integration but also on performance, long considered the home turf of IBM and Oracle.

Could adopting next-generation hardware infrastructure be the key to reclaiming their technological leadership?

NVIDIA, now the most dominant hardware vendor in the AI era, has spent years pushing GPU acceleration for analytics workloads and is beginning to gain traction. Meanwhile, startups like NeuroBlade are bringing purpose-built analytics accelerators to market, already showcasing order-of-magnitude performance and cost savings. Purpose-built solutions like NeuroBlade represent the critical leap forward in data analytics infrastructure.

It’s becoming clear that the first to embrace these next-generation infrastructures will lead the next performance wave—and with it, gain the strategic advantage needed to grow market share and define the future of analytics.

The cloud-native analytics landscape is no longer led by traditional enterprise vendors. Today, platforms like Databricks, Snowflake, and cloud-native offerings from Google (BigQuery), Amazon (Redshift), and Microsoft (Azure Synapse) dominate the conversation. In contrast, IBM and Oracle—once synonymous with enterprise data—have seen their influence wane as customers prioritize agility, cost-efficiency, and ML-readiness.

But the story isn’t finished.

IBM, with the launch of Watsonx.data, and Oracle, through its Autonomous Database and OCI stack, are showing signs of reinvention. The real question is: can they go beyond catching up—can they lead again?

To do so, both must acknowledge a fundamental shift: the most influential voices in technology decisions today are no longer just CIOs. They are data engineers, analysts, ML practitioners, and architects who demand open platforms, real-time performance, and seamless cloud integration. This is where Snowflake wins with elastic compute and simplicity, and where Databricks thrives with its Photon engine and ML-native approach.

Of course, IBM and Oracle must continue to simplify their platforms and deliver seamless multi-cloud and hybrid experiences. These are no longer differentiators—they’re the cost of entry. And they must do all of this while competing with cloud service providers who have virtually unlimited engineering capacity.

To truly leapfrog the competition, IBM and Oracle must rebuild credibility with developers, reshape the conversation, and deliver clear, measurable results. That requires focusing on what has always been the foundation of data infrastructure leadership: performance and cost-efficiency.

Radical Performance and Cost Efficiency Both companies have ceded ground on the axis that once defined them: price-performance. Today, the speed at which queries execute—and the cost per query—are dictated by players who built from scratch with cloud-native assumptions.

IBM’s internal claim that Watsonx.data matches Databricks Photon performance at 60% of the cost is a strong signal, but one that must be backed by third-party validation and real-world benchmarks.

There are two primary paths to performance improvement: better query planning and faster query execution. IBM and Oracle have deep expertise in the former. But Databricks, Snowflake, and the cloud service providers have invested heavily in the latter—an area IBM and Oracle once led, but now lag behind.

Reinventing Query Execution Infrastructure Here lies the best opportunity to retake the lead.

Rather than iterating on legacy execution engines, IBM and Oracle could define a new architecture optimized for accelerated analytics. Technologies like NVIDIA GPUs—which claim up to 100x performance gains for certain workloads—are already being explored and integrated at some of the cloud service providers, and also at both IBM and Oracle. The high costs of GPUs don’t always justify the performance gains, leading to some initiatives like trying to re-purpose old GPUs. So it seems the future lies within customized solutions that were designed with the workload of analytics in mind.

One such example is NeuroBlade’s analytics accelerators, which promise 10x gains at lower cost and are already being explored and piloted by cloud service providers. While there are several other startup companies in the field, NeuroBlade is one of the most advanced solutions currently in deployment. We certainly believe this is just the beginning and additional solutions will come to market soon.

If IBM or Oracle are first to industrialize this approach—not as an add-on, but as a native foundation—they can:

• Regain developer trust by delivering something truly new, open, and performant—not just enterprise-compliant.
• Reposition themselves as innovation leaders in performance, a narrative that will resonate with practitioners, not just procurement teams.

While competitors are still tuning their query planners, IBM or Oracle could redefine the compute layer itself—and with it, shift the momentum of the entire market.

IBM and Oracle are not startups. But they can still operate with a challenger’s urgency. With bold investments in modern execution infrastructure, hardware integration, and developer-first design, they can reestablish themselves as relevant—and respected—leaders.

To lead again, they can’t just build. They must out-innovate.