Problem definition: Artificial intelligence (AI) is rapidly transforming the research and practice of supply chain management. Yet its impact depends on how effectively it is integrated with the theories, methods, and fundamental principles of operations management (OM), which must also evolve to account for the informational, incentive, and institutional changes brought by AI. The OM community has an important role and responsibility to lead in shaping not only how AI transforms supply chains but also how the supply chains that enable AI are designed to be sustainable, resilient, and equitable. Methodology/results: This vision statement organizes the discussion around five layers of the interaction between AI and supply chain management: intelligence, execution, strategy, human, and infrastructure. It synthesizes recent research and industry practice to show how AI enhances forecasting, planning, decision making, risk management, and human-machine collaboration and also examines the supply chains that support AI. Finally, it highlights persistent challenges in data quality, model integration, governance, and workforce adaptation. Managerial implications: Realizing AI's promise in supply chain management requires reliable data and infrastructure, integration of learning and optimization, transparent and explainable decision systems, and a long-term commitment to human-AI collaboration. Together, these elements form the foundation for resilient, adaptive, and trustworthy supply chains in the AI era.

Online Rack Placement in Large-Scale Data Centers: Online Sampling Optimization and Deployment Data centers have grown into major components of global supply chains. This paper develops, deploys, and assesses an optimization algorithm to improve how large-scale data centers place incoming server racks dynamically while balancing space, power, cooling, and reliability constraints. Poor placement decisions can leave valuable resources stranded, leading to high costs and reduced operating resilience. This paper formulates a large-scale online discrete optimization model and develops a new online sampling optimization (OSO) algorithm that anticipates future demand by repeatedly simulating future arrivals and reoptimizing decisions over time. Theoretical results provide performance guarantees, and computational results show its benefits against state-of-the-art reoptimization methods. The system was implemented as a decision-support tool and deployed across Microsoft’s global fleet of data centers. Using postdeployment data, the paper shows that adoption of the tool reduced power stranding by one to three percentage points. At Microsoft’s scale, these improvements translate into substantial financial savings and meaningful reductions in greenhouse gas emissions, demonstrating the real-world impact of optimization in cloud infrastructure management.

Flow optimization is fundamental to computer science, and particularly to computer networks. A prominent example is flow optimization in the global “backbone networks” that interconnect service providers’ datacenters. We tackle a major gap between theory and practice: While in theoretical models upcoming traffic demands are typically known, in real-world networks such information is rarely available a priori. In practice, addressing this gap often involves predicting upcoming demands and optimizing for these. Using data from production networks, we show that this approach can produce solutions that deviate significantly from the optimum. We propose a novel approach: leveraging empirical data to directly learn flow configurations that deliver robustly high performance, bypassing the need for explicit demand prediction. We prove the optimality of our methodology. We further show that by building on recent advances in large-scale optimization and deep learning, our approach enables efficient training on extensive data, picking out intricate patterns in real-world traffic. Through extensive empirical evaluation, we demonstrate that our approach significantly outperforms the state of the art in terms of both solution quality and online runtimes.

Problem definition: Cloud computing is a multibillion-dollar business that draws substantial capital investments from large companies such as Amazon, Microsoft, and Google. Large cloud providers need to accommodate the growing demand for computing resources while avoiding unnecessary overprovisioning of hardware and operational costs. The underlying decision processes are challenging, as they involve long-term hardware and infrastructure investments under future demand uncertainty. In this paper, we introduce the cloud server deployment problem. One important aspect of the problem is that the infrastructure preparation work has to be planned for before server deployments can take place. Furthermore, a combination of temporal constraints has to be considered together with a variety of physical constraints. Methodology/results: We formulate the underlying optimization problem as a two-stage stochastic program. After carefully examining the demand data and on-the-ground deployment operations, we distill two structural properties on deployment throughput constraints and provide tightness results on a convex relaxation of the second stage. Based on that, we develop efficient cutting-plane methods that exploit the special structure of the problem and can accommodate different risk measures. We test our algorithms with real production traces from Microsoft Azure and demonstrate sizeable cost reductions. We show empirically that the algorithms remain optimal even when the two properties are not fully satisfied. Managerial implications: Cloud supply chain operations were largely executed manually due to their complexity and dynamic nature. In this paper, we show that the key decision processes can be systematically optimized. In particular, we demonstrate that accounting for the stochastic nature of demands results in substantial cost reductions in cloud server deployments. Another benefit of our stochastic optimization approach is the ability to seamlessly integrate configurable risk preferences of cloud providers.

In virtual machine (VM) allocation systems, caching repetitive and similar VM allocation requests and associated resolution rules is crucial for reducing computational costs and meeting strict latency requirements. While modern allocation systems distribute requests among multiple allocator agents and use caching to improve performance, current schedulers often neglect the cache state and latency considerations when assigning each new request to an agent. Due to the high variance in costs of cache hits and misses and the associated processing overheads of updating the caches, simple load-balancing and cache-aware mechanisms result in high latencies. We introduce Kamino, a high-performance, latency-driven and cache-aware request scheduling system aimed at minimizing end-to-end latencies. Kamino employs a novel scheduling algorithm grounded in theory which uses partial indicators from the cache state to assign each new request to the agent with the lowest estimated latency. Evaluation of Kamino using a high-fidelity simulator on large-scale production workloads shows a 42% reduction in average request latencies. Our deployment of Kamino in the control plane of a large public cloud confirms these improvements, with a 33% decrease in cache miss rates and 17% reduction in memory usage.

Supply Chain Management requires addressing a variety of complex decision-making challenges, from sourcing strategies to planning and execution. Over the last few decades, advances in computation and information technologies have enabled the transition from manual, intuition and experience-based decision-making, into more automated and data-driven decisions using a variety of tools that apply optimization techniques. These techniques use mathematical methods to improve decision-making. Unfortunately, business planners and executives still need to spend considerable time and effort to (i) understand and explain the recommendations coming out of these technologies; (ii) analyze various scenarios and answer what-if questions; and (iii) update the mathematical models used in these tools to reflect current business environments. Addressing these challenges requires involving data science teams and/or the technology providers to explain results or make the necessary changes in the technology and hence significantly slows down decision making. Motivated by the recent advances in Large Language Models (LLMs), we report how this disruptive technology can democratize supply chain technology - namely, facilitate the understanding of tools' outcomes, as well as the interaction with supply chain tools without human-in-the-loop. Specifically, we report how we apply LLMs to address the three challenges described above, thus substantially reducing the time to decision from days and weeks to minutes and hours as well as dramatically increasing planners' and executives' productivity and impact.

Mathematical programming – the task of expressing operations and decision-making problems in precise mathematical language – is fundamental across domains, yet remains a skill-intensive process requiring operations research expertise. Recent advances in large language models for complex reasoning have spurred interest in automating this task, translating natural language into executable optimization models. Current approaches, however, achieve limited accuracy, hindered by scarce and noisy training data without leveraging domain knowledge. In this work, we systematically integrate optimization expertise to improve formulation accuracy for mixed-integer linear programming, a key family of mathematical programs. Our OptiMind framework leverages semi-automated, class-based error analysis to guide both training and inference, explicitly preventing common mistakes within each optimization class. Our resulting fine-tuned LLM significantly improves formulation accuracy by 20.7

A common challenge in real-time operations is deciding whether to re-solve an optimization problem or continue using an existing solution. While modern data platforms may collect information at high frequencies, many real-time operations require repeatedly solving computationally intensive optimization problems formulated as Mixed-Integer Linear Programs (MILPs). Determining when to re-solve is, therefore, an economically important question. This problem poses several challenges: 1) How to characterize solution optimality and solving cost; 2) How to detect environmental changes and select beneficial samples for solving the MILP; 3) Given the large time horizon and non-MDP structure, vanilla reinforcement learning (RL) methods are not directly applicable and tend to suffer from value function explosion. Existing literature largely focuses on heuristics, low-data settings, and smooth objectives, with little focus on common NP-hard MILPs. We propose a framework called Proximal Policy Optimization with Change Point Detection (POC), which systematically offers a solution for balancing performance and cost when deciding appropriate re-solving times. Theoretically, we establish the relationship between the number of re-solves and the re-solving cost. To test our framework, we assemble eight synthetic and real-world datasets, and show that POC consistently outperforms existing baselines by 2

This paper optimizes the configuration of large-scale data centers toward cost-effective, reliable and sustainable cloud supply chains. The problem involves placing incoming racks of servers within a data center to maximize demand coverage given space, power and cooling restrictions. We formulate an online integer optimization model to support rack placement decisions. We propose a tractable online sampling optimization (OSO) approach to multi-stage stochastic optimization, which approximates unknown parameters with a sample path and re-optimizes decisions dynamically. We prove that OSO achieves a strong competitive ratio in canonical online resource allocation problems and sublinear regret in the online batched bin packing problem. Theoretical and computational results show it can outperform mean-based certainty-equivalent resolving heuristics. Our algorithm has been packaged into a software solution deployed across Microsoft's data centers, contributing an interactive decision-making process at the human-machine interface. Using deployment data, econometric tests suggest that adoption of the solution has a negative and statistically significant impact on power stranding, estimated at 1-3 percentage point. At the scale of cloud computing, these improvements in data center performance result in significant cost savings and environmental benefits.

Mixed Integer Linear Programming (MILP) is essential for modeling complex decision-making problems but faces challenges in computational tractability and interpretability. Current deep learning approaches for MILP focus on specific problem classes and do not generalize to unseen classes. To address this shortcoming, we take a foundation model training approach, where we train a single deep learning model on a diverse set of MILP problems to generalize across problem classes. As existing datasets for MILP lack diversity and volume, we introduce MILP-Evolve, a novel LLM-based evolutionary framework that is capable of generating a large set of diverse MILP classes with an unlimited amount of instances. We study our methodology on three key learning tasks that capture diverse aspects of MILP: (1) integrality gap prediction, (2) learning to branch, and (3) a new task of aligning MILP instances with natural language descriptions. Our empirical results show that models trained on the data generated by MILP-Evolve achieve significant improvements on unseen problems, including MIPLIB benchmarks. Our work highlights the potential of moving towards a foundation model approach for MILP that can generalize to a broad range of MILP problem classes. Our code and data are publicly available at https://github.com/microsoft/OptiGuide.

We study the efficacy of Small Language Models (SLMs) in facilitating application usage through natural language interactions. Our focus here is on a particular internal application used in Microsoft for cloud supply chain fulfilment. Our experiments show that small models can outperform much larger ones in terms of both accuracy and running time, even when fine-tuned on small datasets. Alongside these results, we also highlight SLM-based system design considerations.

We explore a new design point for traffic engineering on wide-area networks (WANs): directly optimizing traffic flow on the WAN using only historical data about traffic demands. Doing so obviates the need to explicitly estimate, or predict, future demands. Our method, which utilizes stochastic optimization, provably converges to the global optimum in well-studied theoretical models. We employ deep learning to scale to largeWANs and real-world traffic. Our extensive empirical evaluation on real-world traffic and network topologies establishes that our approach's TE quality almost matches that of an (infeasible) omniscient oracle, outperforming previously proposed approaches, and also substantially lowers runtimes.

Supply chain operations traditionally involve a variety of complex decision making problems. Over the last few decades, supply chains greatly benefited from advances in computation, which allowed the transition from manual processing to automation and cost-effective optimization. Nonetheless, business operators still need to spend substantial efforts in explaining and interpreting the optimization outcomes to stakeholders. Motivated by the recent advances in Large Language Models (LLMs), we study how this disruptive technology can help bridge the gap between supply chain automation and human comprehension and trust thereof. We design OptiGuide -- a framework that accepts as input queries in plain text, and outputs insights about the underlying optimization outcomes. Our framework does not forgo the state-of-the-art combinatorial optimization technology, but rather leverages it to quantitatively answer what-if scenarios (e.g., how would the cost change if we used supplier B instead of supplier A for a given demand?). Importantly, our design does not require sending proprietary data over to LLMs, which can be a privacy concern in some circumstances. We demonstrate the effectiveness of our framework on a real server placement scenario within Microsoft's cloud supply chain. Along the way, we develop a general evaluation benchmark, which can be used to evaluate the accuracy of the LLM output in other scenarios.

Routing is, arguably, the most fundamental task in computer networking, and the most extensively studied one. A key challenge for routing in real-world environments is the need to contend with uncertainty about future traffic demands. We present a new approach to routing under demand uncertainty: tackling this challenge as stochastic optimization, and employing deep learning to learn complex patterns in traffic demands. We show that our method provably converges to the global optimum in well-studied theoretical models of multicommodity flow. We exemplify the practical usefulness of our approach by zooming in on the real-world challenge of traffic engineering (TE) on wide-area networks (WANs). Our extensive empirical evaluation on real-world traffic and network topologies establishes that our approach's TE quality almost matches that of an (infeasible) omniscient oracle, outperforming previously proposed approaches, and also substantially lowers runtimes.

A main challenge faced by cloud providers is to ensure that they are ready to accommodate the growing demand for compute resources. Towards that goal, providers need to deploy cloud servers agilely for uncertain future demand under many practical business constraints while avoiding unnecessarily large operational costs. In this paper, we introduce the cloud server deployment problem. One important aspect of the problem is that the infrastructure preparation work has to be planned for before server deployments can take place. Furthermore, a combination of temporal constraints (e.g., projected dock dates) has to be considered together with a variety of physical constraints (e.g., hardware compatibility and daily deployment throughput at the data centers). While the problem shares similarities with other supply chain problems, its collection of characteristics requires new solutions that explicitly account for demand uncertainty. We formulate the underlying optimization problem as a two-stage stochastic program, and develop efficient and exact Benders-type algorithms that exploit the special structure of the second stage problem while accommodating different risk measures. We test our proposed algorithms with real production traces from Microsoft Azure, and demonstrate their effectiveness in cost reductions.

Our analysis of a large public cloud ML training service shows that resources remain unused likely because users statically (over-)allocate resources for their jobs given a desire for predictable performance, and state-of-the-art schedulers do not exploit idle resources lest they slow down some jobs excessively. We consider if an anticipatory scheduler, which schedules based on predictions of future job arrivals and durations, can improve over the state-of-the-art. We find that realizing gains from anticipation requires dealing effectively with prediction errors, and even the best predictors have errors that do not conform to simple models (such as bounded or i.i.d. error). We devise a novel anticipatory scheduler called SIA that is robust to such errors. On real workloads, SIA reduces job latency by an average of 2.83× over the current production scheduler, while reducing the likelihood of job slowdowns by orders of magnitude relative to schedulers that naïvely share resources.

The infinite capacity of cloud computing is an illusion: in reality, cloud providers cannot always have enough capacity of the right type, in the right place, at the right time to meet all demand. Consequently, cloud providers need to implement admission-control policies to ensure accepted capacity requests experience high availability. However, admission control in the public cloud is hard due to dynamic changes in both supply and demand: hardware might become unavailable, and actual VM consumption could vary for a variety of reasons including tenant scale-outs and fulfillment of VM reservations made by customers ahead of time. In this paper, we design and implement Kerveros, a flexible admission-control system that has three desired properties: i) high computational scalability to handle a large inventory, ii) accurate capacity provisioning for high VM availability, and iii) good packing efficiency to optimize resource usage. To achieve this, Kerveros uses novel bookkeeping techniques to quickly estimate the capacity available for incoming VM requests. Our system has been deployed in Microsoft Azure. Results from both simulations and production confirm that Kerveros achieves more than four nines of availability while sustaining request processing latencies of a few milliseconds.

Oversubscription is an essential cost management strategy for cloud database providers, and its importance is magnified by the emerging paradigm of serverless databases. In contrast to general purpose techniques used for oversubscription in hypervisors, operating systems and cluster managers, we develop techniques that leverage our understanding of how DBMSs use resources and how resource allocations impact database performance. Our techniques are designed to flexibly redistribute resources across database tenants at the node and cluster levels with low overhead. We have implemented our techniques in a commercial cloud database service: Azure SQL Database. Experiments using microbenchmarks, industry-standard benchmarks and real-world resource usage traces show that using our approach, it is possible to tightly control the impact on database performance even with a relatively high degree of oversubscription.