Supply Chain Management: AI, Spend Control, and Sustainability

AI procurement is moving from experimental analytics to a near-term operating priority because the upside and leakage are now both material. According to www.tellius.com, Gartner forecasts supply chain management software with agentic AI will grow from less than $2 billion in 2025 to $53 billion by 2030, implying roughly 94% compound annual growth from 2025 to 2030. That pace signals a market shift: procurement teams are no longer just asking whether AI can classify spend or summarize contracts, but whether it can actively surface savings opportunities, flag buying deviations, and support faster decisions across sourcing, purchasing, and supplier management. The urgency is also financial. www.tellius.com reports that The Hackett Group's User Experience and Maverick Spend Study found organizations can lose up to 16% of negotiated savings to off-contract maverick spend. The same source cites recent research putting preventable post-signature leakage at roughly 11% of contract value from maverick buying, missed rebates, and unclaimed volume discounts. In that context, AI procurement software is becoming a control layer for protecting negotiated value after contracts are signed, not only a tool for finding savings before sourcing events. External operating pressure adds another reason to act now. www.traxtech.com reports that energy inefficiency in supply chain operations can raise operating costs and pressure margins, while supply chains tied to high-carbon-intensity regions face growing scrutiny from customers and trading partners as sustainability expectations rise. Together, these cost, compliance, and margin pressures make better spend intelligence and automated exception detection increasingly important for procurement leaders. One major trend behind this shift is the movement of agentic AI from procurement experimentation toward a major software spending category. According to www.tellius.com, Gartner forecasts supply chain management software with agentic AI will grow from less than $2 billion in 2025 to $53 billion by 2030, representing a roughly 94% compound annual growth rate over that period. That scale of projected growth signals a shift in how procurement and supply chain teams may evaluate AI: not only as analytics assistance, but as software capable of taking on more autonomous, workflow-oriented tasks. The practical implication is that agentic AI is likely to develop around existing procurement stacks rather than replace them outright. www.tellius.com reports that most enterprises already run a combination of procurement tools, including SAP Ariba or Coupa for buy-and-pay workflows, Tipalti or Stampli for invoice capture and payment, and Tellius as a spend-intelligence layer. In that environment, the trend is toward AI that can operate across fragmented systems, surface opportunities, and support action without requiring companies to rebuild procurement from scratch. For buyers, this makes integration and workflow fit central to the agentic AI conversation. The projected market expansion creates pressure to distinguish between tools that simply add AI features and platforms that can connect spend intelligence with the systems where purchasing, invoicing, and payments already happen. As investment accelerates, procurement leaders will need to assess whether agentic AI capabilities can work across their current mix of applications and whether they improve the visibility and execution gaps that arise when spend data, supplier activity, and payment workflows sit in separate tools. A second major shift is that spend intelligence is moving from after-the-fact reporting to pre-payment loss prevention. Procurement AI is being evaluated less as a dashboard layer and more as a financial control that sits directly over procure-to-pay activity. According to www.tellius.com, spend intelligence uses AI to validate procurement transactions against the relevant contract, testing invoice lines for overbilling, duplicate payments, off-contract spend, and unclaimed rebates so losses can be recovered or prevented before the payment window closes. That distinction matters because the leakage problem often appears after sourcing teams have already negotiated the savings. www.tellius.com reports that organizations can lose up to 16% of negotiated savings to off-contract maverick spend, based on The Hackett Group's User Experience and Maverick Spend Study. The same source cites recent research putting preventable post-signature leakage at roughly 11% of contract value from maverick buying, missed rebates, and unclaimed volume discounts. This is changing the role of AI procurement software. Instead of replacing procurement suites, the spend intelligence layer reads invoices, contracts, and payments from existing systems, then flags what should not be paid. In that model, AI is not primarily running approvals or storing transactions; it is checking whether the transaction matches the commercial terms the business already negotiated. For buyers, the practical trend is a move toward continuous contract-to-invoice validation. The most relevant question is no longer only whether a platform can analyze spend categories, but whether it can detect billing above terms, duplicate payments, off-contract purchases, missed rebates, and volume-discount gaps early enough to stop cash leakage before payment is released. A third trend is that energy visibility is becoming part of supply chain sustainability. Energy use is moving from a facilities issue to a supply chain design issue, especially in markets where grid reliability and clean-power access are uneven. According to www.traxtech.com, South African businesses are being called to reassess how their supply chains consume energy and generate carbon emissions. That means sustainability programs can no longer focus only on visible changes such as recycling or packaging; they also need to account for the power required to run warehouses, transport networks, production assets, backup systems, and digital supply chain tools. The pressure is especially practical in South Africa, where supply chain sustainability is shaped by energy infrastructure reliability, grid dependency, and access to clean power alternatives. When grid challenges force operations to rely on diesel generators and backup power systems, companies can face both higher costs and higher carbon impacts. This makes energy resilience and emissions management connected priorities rather than separate workstreams. A related shift is happening in supply chain technology. AI-powered systems can create operational value, but they also consume electricity through data centers. As a result, the sustainability conversation is expanding to include the energy footprint of the analytics, automation, and optimization tools companies deploy. The most mature response is not to avoid digital systems, but to measure their energy implications alongside the savings and efficiencies they generate. Supply chains that are getting ahead are mapping energy consumption at a granular level across operations, processes, and assets. This creates a clearer view of where energy is consumed, where emissions are produced, and where reliability risks can disrupt service. For buyers, the emerging benchmark is whether providers can connect energy data, operational performance, and carbon impact in one decision-making framework. Stargo’s proprietary supply chain data suggests the next wave of procurement AI value will come from document execution and workflow throughput, not classification alone. In one anonymized deployment, Stargo processed 18,000 purchase-order attachments in the first 30 days without increasing back-office headcount. Across Stargo benchmarks, AI-assisted vendor document normalization reduced supplier onboarding cycle time by 11.4 days on average. The implication for supply chain management teams: AI spend and procurement tools should be evaluated not only on anomaly detection, but also on whether they remove vendor-document friction and approval-routing bottlenecks.