Adaptive6 Targets ‘Shadow Waste’ to Turn Cloud Cost Overruns Into an Engineering Problem

Public cloud costs are climbing rapidly, and generative AI is only accelerating that trend. Yet a significant share of enterprise cloud spend is still effectively burned on waste—duplicated, outdated, or inefficient resources that deliver no real business value. A new startup, Adaptive6, has emerged from stealth with the explicit aim of treating this waste not as a budgeting issue, but as a technical defect embedded in code and architecture.

Armed with $44 million in funding and already deployed at enterprises like Ticketmaster, Bayer, and Norstella, Adaptive6 is pitching a model that borrows directly from cybersecurity: continuously scan cloud environments, map issues back to specific code and owners, and remediate them in the engineering workflow.

The cost of cloud ‘shadow waste’ in the generative AI era

Cloud adoption and AI workloads continue to surge. Gartner projects public cloud spend to grow another 21.3% in 2026, while Flexera’s most recent State of the Cloud report estimates as much as 32% of enterprise cloud spend is wasted. That waste spans duplicated code, non-functional services, outdated components, unnecessary scaffolding, and generally inefficient processes.

Generative AI has compounded the problem. Enterprises are leaning on AI-assisted development—what some call “vibe coding” and “agentic swarming”—to ship features faster. Tools and agents such as Claude Code can generate large volumes of application and infrastructure code in a fraction of the time it once took.

But speed has a downside: AI-generated code is rarely optimized for cost. As Adaptive6 co-founder and CEO Aviv Revach notes, most generative models are trained on broad code corpora that did not prioritize cloud efficiency or governance. The result can be sprawling, overprovisioned architectures that functionally work but quietly inflate bills.

For cloud, FinOps, and engineering leaders, this creates a paradox. AI helps teams move faster and build more, but it also introduces a growing layer of hidden inefficiency—what Adaptive6 calls “Shadow Waste”—that traditional cost visibility tools were never designed to catch.

From finance dashboards to engineering ownership

Most first-generation cloud cost management tools were built for finance and procurement: dashboards, forecasts, trend analyses, and budget controls. They can show how much a team spent last month and where, but they typically stop at the point where action is needed. Engineers still need to figure out what to change and how.

Revach argues that this model fundamentally misplaces ownership. Just as CISOs drive security awareness but cannot patch every vulnerability themselves, FinOps leaders can champion cloud efficiency but cannot directly fix inefficient code or misconfigured resources.

“The first generation of tools are sort of trying to help on the financial side of the cloud,” he explained. “They typically deal with the financial aspects of cloud cost… showing you costs going up, costs going down, forecasting, budgeting. But what they don’t really focus on is one of the biggest problems, which is the waste problem.”

Adaptive6 flips this model by reframing cloud waste as an engineering problem. Instead of asking finance teams to interpret line items they cannot influence, the platform is designed to put precise, actionable findings into the hands of the engineers who own the code and infrastructure. The goal is to make cost governance part of everyday development and operations, not just a monthly reconciliation exercise.

Inside Adaptive6’s ‘Cloud Cost Governance and Optimization’ platform

At the center of Adaptive6’s approach is its Cloud Cost Governance and Optimization (CCGO) platform. Rather than relying on agents installed across infrastructure, the system uses standard cloud provider APIs to gain read-only visibility into environments. It scans across major public clouds—AWS, Google Cloud Platform (GCP), and Microsoft Azure—as well as platform services like Databricks and Snowflake and into Kubernetes clusters.

What Adaptive6 is looking for goes beyond idle or oversized instances. The company focuses on “Shadow Waste”: inefficiencies buried in application logic, configuration settings, and architectural choices that rarely surface in simple utilization charts. This could be redundant services, non-optimal data access patterns, or outdated runtime versions that waste compute by running less efficiently.

Revach describes a core capability the company calls “Cloud to Code” technology. The platform correlates problematic cloud resources back to the specific lines of code that spawned them, and to the engineers or teams responsible. This mapping allows the system to push targeted recommendations into existing workflows, such as Jira, Slack, or ServiceNow, rather than forcing teams into a separate interface.

In practice, CCGO operates more like a continuous scanner than a one-time audit. It identifies cost issues, traces them to their source in code and configuration, and then provides remediation guidance or automated fixes that engineers can review and apply. For organizations with complex, multi-cloud estates, this is intended to turn sprawling cost data into concrete engineering tasks.

Concrete examples: from Python versions to LLM throughput

The value of this approach becomes clearer in specific technical examples. One case Revach highlights is the choice of programming language runtime. For teams using Python, upgrading to a more recent version—such as Python 3.12—can bring significant performance improvements. Faster execution means the same workloads can complete more quickly, reducing the compute time and cost required.

Yet most cloud cost optimization practices focus on obvious levers like instance sizing or storage tiers. The version of Python an application uses is rarely treated as a cost variable, even though it can have real financial impact. Adaptive6’s platform is designed to surface such application-level opportunities, not just infrastructure-level adjustments.

Another example involves AI workloads, particularly large language models (LLMs) running on AWS. Here, engineers must decide how much provisioned throughput to commit for inference or training workloads. Under-committing risks performance problems and degraded user experience; over-committing leads to excess capacity that is paid for but not fully used.

Adaptive6 analyzes actual usage patterns for these AI workloads to recommend more precise throughput commitments. This kind of tuning is difficult to do manually at scale and is generally beyond the scope of finance-driven tools that operate at the level of monthly spend rather than per-service behavior.

By examining both configuration and application behavior, the platform aims to capture a spectrum of opportunities: from switching to more efficient runtimes, to revising how specific services are orchestrated, to tightening commitments for AI services that increasingly represent a large share of cloud bills.

Borrowing the cybersecurity playbook

Adaptive6’s methodology is shaped heavily by Revach’s background in cybersecurity, including his experience as a security research team leader in the Israeli Military Intelligence Unit 8200 and as former Head of Strategy at Taboola. In security, the standard pattern is well known: scan for vulnerabilities, map them to affected systems and code, assign ownership, and remediate or block regressions earlier in the lifecycle.

Revach and his team saw parallels between this and cloud cost waste. “We realized this is not a financial problem; it’s an engineering problem,” he said. “We drew on our background in cybersecurity, where to find vulnerabilities, you scan the cloud, identify the issues, map them back to the relevant code, find the responsible developer or engineer, and remediate—or, in some cases, shift left and prevent them altogether… it was obvious that this is exactly what we need to do.”

That analogy extends to how Adaptive6 uses AI. The platform does employ AI to generate remediation scripts and so-called “1-Click Fixes,” but the company is careful to distinguish this from generic coding agents. Because much of today’s AI-generated code is trained on repositories that did not prioritize efficiency, Adaptive6 leans on a dedicated research team—similar in spirit to security research groups—to discover new patterns of cost inefficiency.

This mix of automated analysis and human-driven research is intended to continuously expand the catalog of detectable waste patterns, much like vulnerability databases evolve in cybersecurity. The aim is to keep pace with emerging architectures and new cloud services, including rapidly evolving AI infrastructure.

Early impact with large-scale enterprises

Adaptive6 is not launching purely as a concept; the platform is already deployed at major enterprises such as Ticketmaster, Bayer, and Norstella. According to the company, customers are seeing reductions of 15–35% in total cloud spend after applying the platform’s findings and fixes.

For global organizations with many teams, regions, and business units, decentralizing cost ownership is a recurring challenge. Central FinOps or platform teams can set policies and provide guidance, but they often struggle to enforce efficient practices across dozens or hundreds of engineering squads.

Revach positions this complexity as a core strength for Adaptive6 rather than a barrier. “As complex as it gets with a big organization, that’s exactly our sweet spot,” he noted. In one case, he says, the platform helped identify a single misconfiguration that, once fixed, led to more than $1 million in savings.

Those kinds of outcomes resonate with leaders who are under pressure to fund AI and modernization initiatives without letting cloud budgets spiral. Instead of broad, top-down mandates to “cut 20%,” Adaptive6 is trying to give organizations a more surgical, engineering-led way to reclaim overspend without bluntly slowing innovation.

Shifting left: preventing cloud waste before it deploys

Beyond detecting and fixing existing issues, Adaptive6 also aims to prevent new waste from entering production. The platform includes “shift left” capabilities that integrate into CI/CD pipelines, scanning code changes for potential cost inefficiencies before deployment.

This pre-deployment analysis mirrors how application security tools flag vulnerabilities in pull requests or block builds that introduce critical issues. In the cost context, it can stop expensive architectural mistakes—such as non-optimal resource configurations or inefficient patterns—from ever reaching live environments.

“We detect what’s already wasting money, prevent new inefficiencies before they deploy, and remediate at scale,” Revach summarized. For engineering managers, this promises a feedback loop where cost awareness becomes another standard quality gate alongside security, reliability, and performance.

The broader implication for cloud and FinOps leaders is a shift in where cloud cost management happens. Rather than living solely in spreadsheets and executive dashboards, cost governance moves into version control systems, issue trackers, and developer chat tools. If Adaptive6’s model takes hold, the future of cloud spend optimization may look less like financial reporting and more like continuous engineering hygiene.

With cloud budgets rising and generative AI workloads multiplying, the pressure to reduce waste without stifling innovation will only increase. Adaptive6 is betting that the only sustainable way to do that is to treat cloud waste as an engineering defect—discoverable, traceable, and fixable in code.