Why AI Cost Management Has Become the Next Engineering Crisis
AI cost management is the discipline of tracking, allocating, and optimizing AI-related spending so engineering teams can connect model usage, infrastructure, and tools to real business outcomes in a measurable way. As AI adoption accelerates across software delivery, invoices grow while clarity shrinks. Engineering leaders see rapidly rising token bills, AI agent charges, and cloud line items, but lack a clear view of which workloads create real value. According to Gartner, worldwide AI software spending is expected to be $2.59 trillion in 2026, yet most organizations still treat AI as a black box from a financial perspective. Harness’s 2026 State of Engineering Excellence report highlights the gap: 94% of engineering leaders say the metrics that matter most are missing from their current measurement frameworks, leaving them unable to defend AI budgets or prioritize the highest-ROI projects.
Harness Steps In With Real-Time AI Spending Visibility
To address this visibility gap, Harness has introduced two products focused on AI cost management and engineering ROI tracking: AI DLC Insights and Cloud & AI Cost Management. Together, they aim to give engineering organizations real-time AI budget visibility, mapping every dollar spent to shipped code, production agents, and outcomes. Trevor Stuart, SVP and GM at Harness, describes the shift in priorities: the first wave of AI adoption prioritized getting teams to use tools; the next wave is about proving those tools have a positive impact. The new offerings are built to answer a simple, pressing question for engineering and finance leaders alike: what are we spending on AI, and where is the return? By tying spend directly to development and production metrics, these tools seek to replace anecdotal AI success stories with measurable, auditable results.
AI DLC Insights: Connecting Token Spend to Shipped Software
AI DLC Insights focuses on the development side, where nearly every new line of code is now written with AI assistance from tools such as Claude Code, Cursor, GitHub Copilot, or Windsurf. Until now, token spend for these tools has been disconnected from engineering outcomes: teams could see usage, but not how much AI-generated code shipped or improved delivery speed. AI DLC Insights adds an on-machine agent in the developer environment that records every AI-generated line of code, tracks token costs by model and tool, and connects that spend to pull requests, tickets, and deployments. This creates a detailed view of engineering ROI tracking: adoption across agents, per-developer and per-team attribution, and where AI spending is wasted on abandoned code, bloated prompts, or expensive model choices. It also correlates AI use with ship rate, pull request cycle time, DORA metrics, and incident data to show whether AI is producing faster, better releases.
Cloud & AI Cost Management: Unit Economics for AI in Production
Once AI agents move into production, the financial picture changes: every customer interaction, resolved support ticket, or automated workflow triggers an inference call somewhere in the cloud. Most teams only see this as a growing invoice, without insight into which agents, sessions, or workflows drive that spend or whether it is worth it. Cloud & AI Cost Management extends Harness’s existing cloud cost optimization capabilities to AI infrastructure, tracking spend down to individual requests and tying it to specific agents or production paths. This gives engineering leaders AI spending visibility at the unit level, turning previously opaque inference costs into understandable unit economics. With that view, teams can spot cost inefficiencies, such as overused high-end models or poorly tuned agents, and redirect AI budgets toward the most valuable services, while giving finance and product leaders a clearer justification for continued AI investment.
