NFL and Microsoft Launch AI-First Copilot on Sidelines

The NFL’s long-running technology relationship with Microsoft has officially entered an AI-first phase: a multiyear extension announced in August upgrades the league’s Sideline Viewing System with Copilot-powered tools, expands Azure AI into scouting and operations, and promises richer fan-facing content and productivity gains across 32 clubs. (news.microsoft.com) (cnbc.com)

Background​

The NFL and Microsoft have partnered for more than a decade, beginning with the introduction of Microsoft Surface tablets as the league’s official sideline devices. That hardware relationship evolved into a managed, league-controlled Sideline Viewing System (SVS) that today serves coaches, players, and club staff with high-resolution replay images and game-day telemetry. The latest extension builds on that operational foundation by layering Microsoft Copilot, Azure AI tooling, GitHub Copilot features, and new Surface Copilot+ PCs into the same workflows. (microsoft.com) (news.microsoft.com)
The public announcement frames the move as pragmatic: use natural-language copilots to surface insights faster, not to hand decision-making to algorithms. Microsoft’s press materials and independent reporting emphasize human-in-the-loop controls and explicit prohibitions on autonomous AI play-calling. Still, the upgrade is more than a marketing refresh — it is an operational bet that conversational AI can materially reduce the time between question and insight in one of the highest-pressure professional environments. (news.microsoft.com) (axios.com)

---

What was announced — the essentials​

Sideline and device upgrades​

• The Sideline Viewing System has been upgraded with more than 2,500 Microsoft Surface Copilot+ PCs to equip all 32 NFL clubs, roughly 1,800 players, and over 1,000 coaches and football staff with real-time game data and analysis tools. (news.microsoft.com) (cnbc.com)
• The new devices are described as AI-enabled, 5G-capable Surface units (reported in league interviews as Surface Pro 11-class hardware in some briefings) that include on-device AI acceleration and tighter integration with Azure-hosted inference and video services. (sportsbusinessjournal.com)

Copilot features on-field and in the booth​

• Natural-language Copilot assistants on sideline tablets and booth dashboards let coaches and analysts query plays, filter film by down/distance/scoring events, and request clip pulls or comparative statistics in seconds. The toolset emphasizes retrieval and synthesis rather than prescriptive play-calling. (news.microsoft.com) (axios.com)
• A GitHub Copilot–style filtering feature is being shipped to allow fast, ad hoc filtering of plays and key moments — a practical efficiency for moment-driven decisions like challenge reviews or late-game substitutions. (news.microsoft.com)

Scouting, Combine, and analytics​

• The NFL Combine App and scouting suites were piloted with Azure OpenAI–powered assistants at the 2025 Combine, giving scouts interactive, natural-language comparisons across prospect metrics and instant highlight compilations for more than 300 prospects. Microsoft and the league position this as an acceleration of the scouting workflow, enabling iterative “ask and refine” analysis. (news.microsoft.com) (news.microsoft.com)

Operations, front office, and fan experiences​

• The partnership expands Azure’s footprint for game-day operations dashboards that catalog incidents (weather delays, equipment faults), support game-operations managers, and help clubs deploy AI agents across business functions (ticketing, salary-cap planning, HR workflows). Microsoft also highlights the potential to scale personalized fan content: rapid highlight reels, post-game summaries, and targeted marketing powered by behavioral and transactional signals. (news.microsoft.com) (sportsbusinessjournal.com)

---

The Tampa Bay Buccaneers example: marketing meets AI​

Individual clubs are already experimenting with Copilot beyond coaching workflows. The Tampa Bay Buccaneers used Microsoft’s AI tooling to mine historical archives and shape marketing and in-stadium activations tied to their "Creamsicle" retro celebrations and the Spirit of 1979 events. Those use cases — crafting themed content, fan-facing narratives, and targeted promotions — preview how Copilot can accelerate content creation for clubs and sponsors. (buccaneers.com) (mediapost.com)
These trials show how the same AI plumbing that helps a coach find the right clip can be repurposed to generate social posts, compile highlight reels, or create merch-driven offers — lowering production time and expanding personalization at scale.

---

Why this phase matters: technical and operational context​

The announcement is important because it moves generative conversational AI from optional analytics tooling into mission-critical, time-sensitive workflows. Coaching and scouting operate under severe time constraints; shaving seconds from information retrieval can change in-game communication and substitution choices.
Technically, Microsoft and the NFL are combining:

• Edge + cloud design: on-prem/edge caches in stadium Sideline Communications Centers for deterministic latency plus cloud-hosted model inference for heavier analytics. (mediapost.com)
• Azure OpenAI / Copilot models: for natural-language understanding and synthesized answers tied to structured telemetry and video metadata. (news.microsoft.com)
• Managed device fleet and hardened SVS: league-controlled device images, wipe/collection policies, and private Windows servers to preserve parity and secure sensitive data. (news.microsoft.com)

These design choices acknowledge the operational realities of NFL stadiums — mixed venues, variable connectivity, and very low tolerance for downtime. The real challenge will be delivering consistent, auditable responses under peak stadium loads while protecting competitive equity. (mediapost.com)

---

Strengths — what the NFL and clubs stand to gain​

• Speed-to-insight: Natural-language Copilot interfaces collapse hours of manual spreadsheet work into conversational queries, making scouting and prep faster and more iterative. This is valuable for halftime adjustments, challenge reviews, or late-game personnel decisions. (axios.com)
• Operational continuity: Microsoft’s existing SVS footprint and years of device management reduce integration risk. Upgrading an embedded platform is less disruptive than introducing an entirely new vendor stack. (microsoft.com)
• Scalability and security posture: Azure’s enterprise SLAs, global regions, and security certifications give the league a familiar foundation for high-volume live events and controlled data governance. Centralizing telemetry on Azure simplifies backups, disaster recovery, and identity management for clubs. (news.microsoft.com)
• Cross-department reuse: The same Copilot and Azure tooling can be applied across scouting, broadcast, marketing, ticketing, and HR — creating a shared data and AI fabric that reduces duplication and accelerates product development. (sportsbusinessjournal.com)
• Faster content production: Clubs and the league will likely reduce time and cost to produce post-game summaries, highlight reels, and personalized content — an attractive monetization path for broadcasters and direct-to-fan products. (news.microsoft.com)

---

Risks and limits — operational, legal, and ethical​

The promises are notable, but the margin for error is narrow. Key risks include:

• Vendor concentration and systemic exposure: Centralizing mission-critical tooling with a single provider increases the impact of any outage or policy change. A cloud incident or service degradation at peak could affect multiple clubs simultaneously. Mitigation: multi-region failover, on-prem caches, and well-rehearsed incident-response playbooks. (mediapost.com)
• Explainability and hallucinations: Generative models can synthesize compelling outputs that may omit provenance or misattribute stats. In high-stakes contexts, any hallucinated insight could mislead analysts. Mitigation: per-answer provenance metadata, confidence scores, and mandatory human verification for operational use. (mediapost.com)
• Competitive fairness and parity: The league must preserve competitive equity across clubs. If some teams gain privileged data pipelines or superior model tuning, competitive balance could shift. The NFL’s device parity and managed images are a start; continued transparency — including shared baseline tools or disclaimers — is essential. (news.microsoft.com)
• Player privacy and labor implications: AI-driven scouting or injury analytics could affect contract negotiations and player market value. Use of sensitive biometric or medical data must be governed by explicit retention, consent, and union-negotiated terms. Expect legal scrutiny and potential collective-bargaining discussions. (news.microsoft.com)
• Operational reliability under stadium constraints: Stadium networks are heterogeneous; deterministic latency is non-negotiable for many in-game use cases. Edge caching, offline fallbacks, and deterministic SLAs must be validated under real-world load tests. (mediapost.com)

---

Where claims need independent verification​

Several public numbers and technical details should be treated as directional until independently audited:

• The exact inventory of deployed devices can fluctuate with procurement and replacement cycles; Microsoft’s press release cites “more than 2,500 Surface Copilot+ PCs,” a figure corroborated by major outlets but worth treating as an approximate snapshot rather than a fixed count. (news.microsoft.com) (cnbc.com)
• Precise model versions, training data provenance, and runtime SLAs for Copilot responses are not fully disclosed in public materials. These are the most load-bearing technical details for trust and auditability and should be published (or independently audited) as the rollout matures. (mediapost.com)
• The operational footprint of GitHub Copilot features (e.g., exactly how play-filtering queries are translated into deterministic clip pulls) will require hands-on testing to confirm latency and correctness under game-day conditions. (sportsbusinessjournal.com)

When public statements are imprecise, teams and regulators should demand audit-ready artifacts: model-version IDs, per-answer provenance headers, and post-incident root-cause reports.

---

Practical checklist — what teams and league IT should insist on​

• Publish a staged rollout calendar with independent operational readiness reviews before activating Copilot features at regular-season games. (mediapost.com)
• Surface provenance metadata for every Copilot response (model version, timestamp, data sources, and confidence). (mediapost.com)
• Require multi-region failover and edge-cache validation under simulated stadium loads, and run red-team stress tests. (mediapost.com)
• Negotiate explicit data governance and retention terms with player representation (NFLPA), including opt-out and audit rights. (news.microsoft.com)
• Invest in mandatory role-based training for coaches, scouts, and analysts to interpret AI outputs correctly and follow escalation protocols when outputs are ambiguous. (sportsbusinessjournal.com)

---

Governance, legality, and public trust​

The public messaging from the NFL and Microsoft includes guardrails — “AI to assist, not replace” — but good governance requires more than slogans. Effective regulation in this domain should include:

• Transparency: public dashboards or restricted-access reports showing model versions in use and anonymized usage logs.
• Auditability: independent third-party operational audits of latency SLAs, failover behavior, and access logs.
• Privacy-first contracts: clear clauses about what player- and fan-related data can be used to train models, retention windows, and cross-border transfer safeguards.
• Labor protections: negotiation with player unions around analytics used in personnel decisions and explicit protections if analytics feeds into contract or medical decisions.

Without these measures, the league risks erosion of trust — both with players (over privacy and market impacts) and with fans (over opaque personalization and monetization practices). (mediapost.com)

---

The competitive landscape: why Microsoft​

Microsoft’s vertical stack — Surface devices, Windows/servers, Azure cloud, and Copilot models — gives it an execution advantage: the company can deliver integrated device-to-cloud workflows with single-vendor tooling, support, and SLAs. That integration accelerates rollout speed and reduces integration complexity for teams. However, that same verticality concentrates operational dependency and negotiation leverage in one partner, raising long-term vendor-lock-in questions that other cloud providers and sports-tech firms will seek to exploit. (microsoft.com) (sportsbusinessjournal.com)
Leagues and clubs should weigh the short-term benefits of a unified platform against the strategic risk of reduced bargaining power and the difficulty of migrating away if requirements change.

---

Fan impact and commercialization potential​

Expect near-term consumer-facing rollouts built from the same infrastructure:

• Faster highlight generation: Copilot-driven clip curation that surfaces key moments personalized to fan tastes.
• Interactive Q&A: natural-language query features inside team or league apps for historical stats and situational analytics.
• Targeted offers: AI-driven segmentation for ticketing and merchandise, delivering more relevant promotions to fans. (news.microsoft.com)

These features present clear monetization paths but also require careful privacy safeguards and user transparency so personalization doesn’t feel intrusive or discriminatory. (mediapost.com)

---

Summing up — pragmatic optimism, tight guardrails​

The NFL–Microsoft extension is a strategically consistent evolution of a decade-long relationship. The immediate operational upsides — faster access to film and stats, more iterative scouting, and faster content production — are real and already validated in early pilots like the 2025 Combine and club marketing experiments. (news.microsoft.com) (buccaneers.com)
But the margin for error is narrow. Success depends on disciplined engineering for latency and redundancy, transparent provenance and audit trails for Copilot outputs, legally robust data governance with player representation, and staged rollouts that prioritize reliability over flashy capability. In short, Copilot can be an incremental multiplier for human expertise — if the league and its partners embed explainability, independent auditability, and contingency planning into the deployment from day one. (mediapost.com)
The coming preseason and early-season activations will be the most consequential weeks to watch: they will show whether Copilot is a dependable assistant on game day or an operational experiment that needs further hardening. The technical promise is large; the operational, legal, and ethical responsibilities are larger.

---

Source: MediaPost NFL Expands Microsoft Copilot Partnership Following Buccaneers' Campaign

 

[ChatGPT]

The NFL has quietly but decisively moved its long-running technology partnership with Microsoft into an AI-first operational phase, extending a multiyear deal that layers Microsoft Copilot, Azure AI, and new Surface Copilot+ devices into sideline workflows, scouting, stadium operations, and fan-facing marketing—an upgrade confirmed by league and vendor statements and widely reported by independent outlets. (news.microsoft.com) (cnbc.com)

Background / Overview​

The NFL and Microsoft have a relationship that began more than a decade ago with the introduction of Microsoft Surface tablets on the sidelines. What started as a hardware-and-marketing program evolved into a league-managed Sideline Viewing System (SVS) that supports replay, telemetry, and game-day workflows. The latest extension formalizes a shift from device sponsorship to a strategic technology integration where generative AI and cloud services become part of mission-critical match‑day tooling.
Public announcements describe three immediate pillars of the extension:

• Copilot-enabled Surface devices and booth dashboards to give coaches and analysts conversational access to plays, personnel groupings, and snap counts.
• Azure OpenAI–driven scouting tools, including a Combine App assistant that delivers real-time, natural-language player comparisons.
• Expanded Azure and edge deployment across stadiums for game-day operations, content production, and backend scalability.

These elements were laid out in Microsoft’s announcement and corroborated by major business and sports press coverage. The league frames the initiative as assistance, not autonomy—AI to accelerate insight while final decisions remain human. (news.microsoft.com, axios.com)

---

What the Extension Actually Includes​

Sideline Viewing System upgrades and Surface Copilot+ rollout​

The NFL’s Sideline Viewing System has been upgraded with a new generation of devices branded as Microsoft Surface Copilot+ PCs, deployed across clubs to bring conversational AI directly into coaches’ hands. Public materials specify more than 2,500 Surface Copilot+ units provisioned to support 32 clubs, roughly 1,800 players, and over 1,000 coaches and football staff. That device-count figure appears consistently in vendor and press statements, but should be treated as an approximate, auditable inventory rather than an immutable contractual detail. (prnewswire.com, news.microsoft.com)
Key capabilities emphasized for the SVS:

• Natural-language queries for play histories, personnel groupings, and snap counts.
• Rapid clip pulls and synthesized statistical summaries for booth-to-sideline collaboration.
• A GitHub Copilot–style filtering tool that lets analysts quickly filter plays by down/distance, penalties, scoring plays, and other criteria to surface relevant moments in seconds. (news.microsoft.com)

Scouting and the NFL Combine: real-time AI assistants​

The NFL piloted an Azure OpenAI–powered assistant within the NFL Combine App to provide scouts and team personnel with near-real-time answers during the Combine. The assistant was built on Azure OpenAI, Azure Cosmos DB, and Azure Container Apps to provide low-latency, structured responses to natural‑language queries, enabling scouts to ask iterative questions such as cross-season comparisons or “show top 40‑yard dash times for players under 6’0”” and receive instant, structured comparisons and highlight reels. This Combine pilot is described in technical customer stories published by Microsoft and independently confirmed by reporting. (microsoft.com, axios.com)

Operations, analytics and fan-facing services​

The partnership extends beyond the sideline to support:

• Game-day operations dashboards for incident cataloging (weather, broadcast faults, equipment issues).
• Business process automation across ticketing, finance, HR, and event management using AI agents.
• Fan engagement and content production, including rapid highlight reels, post-game summaries, and personalized marketing driven by browsing, purchase, and social signals — capabilities some clubs (e.g., the Tampa Bay Buccaneers) already trialed in marketing campaigns. (buccaneers.com)

---

The Tampa Bay Buccaneers Case Study: Marketing Meets Copilot​

The Buccaneers used Copilot in recent club-level marketing and fan-engagement projects—most visibly during the Creamsicle Game / Spirit of 1979 celebration—applying Copilot to mine historical season data, refine creative designs, and shape gameday experiences that resonated with both legacy and younger fans. Club executives described Copilot as helpful in uncovering archival insights and accelerating content creation for promotional campaigns. That real-world club use served as an early example of how Copilot can be applied beyond coaching and scouting into brand activation and merchandise targeting. (buccaneers.com)
What the Buccaneers example demonstrates in practice:

• Faster production of themed assets and timeline-driven communications.
• Data-driven segmentation for targeted offers and lookalike audiences.
• Rapid generation of recap copy and highlight reels for social channels.

This use case confirms a broader pattern: tools built for internal workflows often get repurposed into fan‑facing experiences, creating value but also raising governance questions about data privacy and personalization thresholds.

---

Technical Anatomy: How Microsoft and the NFL Say It Works​

Public technical descriptions paint a hybrid edge+cloud architecture designed to meet the NFL’s stringent latency and reliability needs.

• Azure OpenAI models power natural-language understanding and synthesis for Copilot experiences.
• Azure Cosmos DB and microservices provide fast structured access for player metrics and event metadata.
• Azure Container Apps host scalable APIs for event-day load without heavy infrastructure management.
• Edge caching and on‑prem nodes are used inside stadium Sideline Communications Centers to mitigate network hiccups and ensure deterministic response times for in-game queries. (microsoft.com, news.microsoft.com)

Notably, public materials do not disclose:

• The precise model versions in production.
• The full training or data‑provenance pipelines that influence responses.
• Per-answer provenance or confidence metadata surfaced to end users.

These omissions are consequential: for high-stakes, real-time decisions, transparency about model provenance and runtime SLAs matters as much as raw latency numbers.

---

What This Means for Coaches and Scouts​

The headline benefit is speed-to-insight. In an environment where seconds can alter challenge decisions, substitutions, or halftime game plans, the ability to surface comparative stats and relevant clips in natural language reduces friction in the information pipeline.
Practical coach- and scout-level benefits:

• Quick multi-season comparisons and personnel breakouts during time-limited windows.
• On-demand clip compilation for immediate review in the booth or sideline.
• Iterative hypothesis-and-validate workflows for scouts at events like the Combine.

However, the NFL has been explicit about governance guardrails: the system is designed as a decision-support tool, not an autonomous play-caller. League communications reaffirm human-in-the-loop controls and device parity measures to avoid bespoke competitive advantages. Those guardrails are essential to keep responsibility and liability with coaches and front-office staff. (news.microsoft.com, axios.com)

---

Governance, Fairness, and Player Privacy: The Real Constraints​

Integrating generative AI into mission-critical sports workflows raises several governance priorities that the league must address decisively:

• Explainability & provenance: Every answer affecting tactical or personnel choices should include provenance metadata—what data sources were consulted, the model version used, and a confidence score. Without this, teams risk over‑reliance on opaque summaries.
• Parity & competitive balance: Device images, data access, and model tuning must be standardized across clubs to avoid asymmetric advantages. The league’s device collection and wipe policies must remain robust.
• Data protection & player privacy: Player health, biometric, and scouting data are sensitive. Policies must define retention, sharing, and allowable use cases, and ensure compliance across jurisdictions.
• Auditability & SLAs: Third-party audits and documented runtime SLAs will be necessary to ensure the system behaves predictably during peak game windows.

These are not theoretical. Public reporting and industry analysis repeatedly highlight that centralizing operational tooling with a single cloud provider reduces integration overhead but increases systemic exposure to outages and vendor policy changes. The trade-off between operational continuity and vendor concentration must be actively managed.

---

Risks, Failure Modes, and Practical Mitigations​

• Model hallucination and misleading outputs
• Risk: Generative models can produce plausible but incorrect statements when faced with ambiguous or out-of-distribution queries.
• Mitigation: Surface provenance, require human validation in UI workflows, and employ conservative default behaviors in critical windows (e.g., present raw data alongside any summarization).
• Single-vendor dependency and outage exposure
• Risk: Migrating more workloads to Azure raises the league’s systemic risk in a major Azure outage or policy change.
• Mitigation: Build multi-region, multi-cloud fallback paths and keep hardened on-prem caches for core SVS features that must remain available during cloud interruptions.
• Competitive imbalance
• Risk: Differential access to enriched datasets or bespoke model training could shift competitive balance.
• Mitigation: League-enforced device parity, controlled dataset releases, and centralized model management with transparent audit logs.
• Privacy and compliance lapses
• Risk: Misuse or improper retention of player or fan data could trigger regulatory or union challenges.
• Mitigation: Clear data retention schedules, role-based access controls, and regular compliance audits.
• Usability and reliability in high-pressure contexts
• Risk: Slower-than-expected responses, clumsy UIs, or unreliable device battery/network behavior can create frustration or worse—delayed decision-making.
• Mitigation: Real-world stress testing, pilot phases with rollback plans, and UI design focused on deterministic, predictable outputs.

These mitigations align with guidance from enterprise AI governance frameworks and are consistent with best practices for deploying generative AI in high-stakes operational settings.

---

The Competitive and Market Implications​

Microsoft’s expanded presence in the NFL deepens its lead in the sports-technology vertical. The partnership demonstrates how long-term hardware relationships can morph into strategic, cloud-centric alliances—raising the bar for competitors and locking an entire operational stack (devices, cloud, models) into a single vendor domain.
Market implications:

• For Microsoft: The NFL deal is a visible flagship showing Azure, Copilot, and Surface working together in real-time, high-stakes environments—an attractive proof point for other rights holders and leagues.
• For other cloud vendors: Expect more aggressive positioning from rival clouds and sports-tech providers emphasizing multi-vendor resiliency, open data standards, or niche models tailored for scouting.
• For clubs and vendors: Procurement decisions will increasingly weigh long-term vendor lock-in risks against the immediate operational benefits of integrated platforms.

The strategic play here is clear: sports leagues are lucrative, high‑visibility labs for enterprise AI capabilities—if the work is done without causing public failures.

---

Practical Recommendations (What the NFL, Teams, and IT Leaders Should Do Next)​

• Publish a staged rollout calendar with externalized milestones and third-party operational readiness reviews.
• Require per-response provenance and model-version metadata surfaced in the user interface by default.
• Implement multi-region failover tests and on-prem edge caching validation under simulated stadium loads.
• Standardize device images and enforce parity rules with automated compliance checks and device collection policies.
• Define explicit data-retention policies and privacy guardrails for player, scouting, and fan data, coordinated with union and regulatory stakeholders.
• Commission an independent audit of model behavior and performance after the first in‑season deployment window and publish a summary of findings to build trust.

These steps balance speed with accountability and will materially reduce the risk of operational surprises while preserving the upside of the new capabilities.

---

How Clubs Can Extract Value Without Increasing Risk​

• Use Copilot primarily for retrieval and synthesis (clips, counters, quick comparisons) rather than prescriptive recommendations in decisive moments.
• Build UI guardrails that require a short human confirmation step before any tactical change suggested by Copilot is actioned.
• Train coaches and analysts to interpret model outputs, including reading provenance and confidence indicators.
• Leverage the same infrastructure for off-field efficiency (ticketing, finance, HR) where the governance burden is lower and ROI accrues quickly.

When used as designed—accelerating information flow and reducing the busywork of data wrangling—Copilot can improve efficiency without supplanting human judgment.

---

A Measured Verdict​

The NFL–Microsoft extension is a pragmatic transformation of a decade-long relationship into an explicit AI-first operational partnership. The technical upside is clear: faster access to verified data, rapid clip compilation, and lower friction for scouting and operations. These are not small conveniences—they can compress hours of manual work into seconds and materially improve situational awareness on game day. (microsoft.com, axios.com)
Yet the margin for error remains narrow. Success will depend on:

• Transparent provenance and explainability for model outputs.
• Robust, tested edge+cloud architectures to ensure deterministic latency and uptime.
• League-level governance that enforces parity and protects sensitive player and fan data.

If those guardrails are enforced and independently audited, the partnership could become a blueprint for safe, high-value AI adoption in professional sports. If not, the same tools that amplify insight could amplify systemic failures: opaque decision-making, concentrated vendor risk, or unexpected competitive imbalances. The league’s public language—emphasizing assistance rather than autonomy—is a responsible start; the decisive work will be in the engineering, policies, and audits that follow rollout.

---

Final Thoughts​

This move underscores a broader trend in enterprise IT: platform consolidation around cloud + generative AI, with verticalized user experiences built on long-term device footprints. The NFL is a particularly demanding proving ground—millisecond expectations, high public visibility, and a responsibility for player safety and competitive fairness. The league’s decision to formalize human-in-the-loop rules and invest in edge resiliency is sound. Still, the community of teams, technologists, regulators, and unions must insist on transparency and auditability as the technology matures.
In short: the technology is promising, the initial architecture and guardrails are sensible, and the real test will be how reliably and transparently these systems perform under the pressure of a full in‑season rollout. The next months should produce pilots, SLAs, published audit results, and (crucially) visible evidence that the league is prioritizing explainability and parity as aggressively as it is pursuing capability. (news.microsoft.com, cnbc.com)

---

Source: MediaPost NFL Expands Microsoft Copilot Partnership Following Buccaneers' Campaign