Mogin Law AI Deal Table shows how quickly firms are moving to secure structural advantages as AI markets take shape.
As artificial intelligence markets mature, the most consequential deals are no longer about consumer‑facing applications or standalone models. Instead, mergers, acquisitions, and major investments increasingly target the infrastructure layers that determine who can compete in AI at scale.
Our updated Mogin Law AI Deal Table reflects this dynamic. Many of the most significant transactions are what economists and antitrust scholars have described as “chokepoint deals,” transactions that consolidate control over critical inputs or distribution layers competitors must use. These moves aim to leverage points across the AI tech stack (the layered combination of hardware, software, and infrastructure required to run AI systems).
Economists Susan Athey, The Economics of Technology Professor at the Stanford Graduate School of Business, and Fiona M. Scott Morton, Theodore Nierenberg Professor of Economics at Yale School of Management —both former chief economists of the U.S. Department of Justice’s Antitrust Division—along with Maurice Stucke, a professor of law at the University of Tennessee, and Ariel Ezrachi, a professor of competition law at the University of Oxford, have identified these dynamics as central to emerging competition risks in AI markets.
Key Facts
- The updated Mogin Law AI Deal Table includes more than 200 transactions spanning acquisitions, investments, partnerships, and large financing rounds across the AI ecosystem.
- A substantial share of deal value remains undisclosed, consistent with strategic infrastructure and capability acquisitions rather than traditional revenue‑driven buyouts.
- The concentration of activity in the past two years underscores how quickly firms are moving to secure structural advantages as AI markets take shape.
What Is the AI Tech Stack?
The AI tech stack consists of several interdependent layers. At the foundation is compute (the processing power—chips, servers, and data centers—used to train and run AI models), typically delivered through hyperscale cloud infrastructure (remote data centers and services used to store data and run applications).
Above compute are software layers such as security and observability (software that monitors system performance, reliability, and errors in real time), which have become essential as AI systems operate continuously and autonomously. At the top are applications and enterprise platforms that deliver AI tools to users.
The Organization of Economic Coordination and Development (OECD) – a global policy group comprising nearly 40 countries – and national competition authorities have emphasized that this layered stack is capital‑intensive and highly interdependent, meaning control over one layer can influence competition well beyond that layer alone. (OECD has some great info on their site. Check out their report on AI adoption by individuals and firms.)
What Makes a Deal a “Chokepoint” Deal?
A chokepoint deal is defined less by deal size than by where in the stack it occurs. These transactions strengthen control over bottlenecks (points in a market where limited access can restrict competition downstream), such as access to compute, security telemetry, data platforms, or enterprise users.
Economists Susan Athey and Fiona Scott Morton have warned that AI markets are particularly vulnerable to bottlenecks in “various layers of the AI stack,” especially where incumbents control essential inputs or distribution. Antitrust scholars Maurice Stucke and Ariel Ezrachi similarly caution that competition risks arise when firms “extend their power across layers,” even absent traditional horizontal mergers.
Read our previous post December AI Deals Intensify Competition in Compute, Infrastructure, and Agentic Automation.
March 2026 Update
Transactions announced since our February review reinforce—and materially sharpen—the chokepoint dynamics identified in the Mogin Law AI Deal Table. Recent deals increasingly involve long‑duration commitments and vertical integration across electricity generation, AI‑ready data centers, advanced chips, and optical interconnects. As reflected in transactions involving hyperscalers, infrastructure investors, utilities, and semiconductor suppliers, competitive advantage in AI is now being shaped not only by control over existing capacity, but by the pre‑allocation of future inputs years in advance. Economists and competition authorities have warned that when access to essential inputs such as power, compute, and data movement is locked in through multi‑year arrangements, the resulting foreclosure risks may arise well before traditional market shares are measurable. These developments underscore that AI competition is increasingly governed by infrastructure control and duration, not model performance alone—placing renewed emphasis on antitrust scrutiny of vertical integration and access to essential facilities.
Chokepoint Deals in the AI Era: Recent Examples
Cloud Security and Multicloud Control
One of the largest AI‑era transactions is Google’s $32 billion acquisition of Wiz. Google itself described the deal as accelerating “two large and growing trends in the AI era: improved cloud security and the ability to use multiple clouds,” or multicloud (the use of multiple cloud providers rather than a single vendor).
Cloud security increasingly functions as a control layer for AI deployment, compliance, and monitoring. As competition authorities and economists have noted, when such a layer is controlled by a major cloud provider, the competitive implications extend beyond pricing to switching costs, interoperability, and long‑term dependency across the stack.
Security, Observability, and Data Platforms
Cisco’s $28 billion acquisition of Splunk illustrates how chokepoints can emerge through platform integration. Cisco emphasized that organizations adopting AI require infrastructure, data, security, and observability to function “all together.”
Observability has become mission‑critical for AI systems that must operate continuously and reliably. When observability is integrated with security and networking under a single platform, competition concerns may arise even if no individual market appears concentrated in isolation—a pattern highlighted in OECD analyses of AI infrastructure markets.
A similar dynamic appears in Palo Alto Networks’ $3.35 billion acquisition of Chronosphere. Palo Alto characterized always‑on observability as essential for modern AI data centers. As regulators and scholars have observed, consolidation in such required layers can affect entry and expansion for AI‑native firms that depend on independent monitoring tools.
Enterprise Workflow as Distribution
Workday’s approximately $1.1 billion acquisition of Sana highlights a different chokepoint: distribution (the channels through which AI tools reach users, such as enterprise software platforms). Workday described the combined platform as the “new front door for work,” integrating AI‑powered search, AI agents (software programs that autonomously perform tasks using AI models), and learning directly into enterprise workflows.
Researchers studying AI competition have identified enterprise interfaces as emerging distribution bottlenecks, particularly where access to users and organizational data is mediated by a single platform. When rivals must integrate with that interface to reach users, distribution itself can become a chokepoint.
AI Implementation and Safety as Scarce Inputs
Not all chokepoints are purely technological. Accenture’s acquisition of the AI firm Faculty—listed in the AI Deal Table at $1 billion—reflects consolidation around implementation capability. Accenture emphasized Faculty’s expertise in deploying “safe and secure” AI systems in mission‑critical settings.
In regulated industries, the number of firms capable of responsibly deploying AI at scale may be limited. Scholars and policymakers have increasingly recognized that such talent, governance frameworks, and execution capacity can function as competitive bottlenecks even without platform dominance.
Capital Formation and Scale Effects
Even transactions that are not acquisitions can reinforce chokepoints. The AI Deal Table includes a 2026 Databricks financing round exceeding $7 billion at an approximate $134 billion valuation, which the company stated would accelerate products built for AI agents and enterprise data access.
As the OECD has noted, AI infrastructure is capital‑intensive. Large financing rounds can amplify scale advantages, making it harder for smaller rivals to compete for customers, talent, or compute—even absent exclusionary conduct.
Hyperscaler Chokepoints in National AI Markets
The consolidation reflected in recent AI‑related transactions overlaps directly with competition concerns surrounding hyperscalers—the small group of cloud providers that control a dominant share of large‑scale computing capacity, storage, and AI‑ready infrastructure in national markets.
As the OECD has observed, artificial intelligence depends on capital‑intensive infrastructure inputs—advanced chips, data centers, and cloud platforms—that are increasingly concentrated among a handful of firms. When control over these inputs converges with downstream software, security, and data‑platform acquisitions, the result can be chokepoints that shape who can train, deploy, and scale AI systems at all.
This dynamic is particularly salient in national markets, where access to hyperscale compute is often a prerequisite for meaningful competition, and where switching costs, interoperability constraints, and capacity scarcity can amplify the competitive effects of otherwise discrete transactions.
(Read our previous post Are Clouds Too Sticky? Antitrust Authorities Probe Lock-in Pricing Complaints.)
Agencies Are Explicitly Examining Cloud and AI Infrastructure Power
Recent enforcement activity indicates that regulators are not treating these issues as theoretical.
In January 2025, the Federal Trade Commission issued a staff report examining partnerships and investments between leading cloud service providers—Alphabet, Amazon, and Microsoft—and major AI developers, highlighting potential competition concerns related to access to computing resources, increased switching costs, and the sharing of sensitive technical information.
Separately, U.S. and international antitrust authorities—including the FTC, the Department of Justice, and the OECD—have issued joint statements and analytical reports emphasizing vigilance across the AI infrastructure stack, with particular attention to cloud computing, interoperability, and control over essential inputs. More recently, the FTC has intensified scrutiny of cloud and AI bundling practices in the enterprise software market, signaling that hyperscaler conduct in national markets is squarely within the current enforcement lens.
The European Commission has opened formal market investigations under the Digital Markets Act to assess whether Amazon Web Services and Microsoft Azure should be designated as cloud “gatekeepers,” citing concerns about interoperability barriers, switching costs, and bundling practices that may entrench hyperscaler dominance in AI‑critical infrastructure.
In the United Kingdom, the Competition and Markets Authority concluded in July 2025 that competition in cloud infrastructure services “is not working well,” and has recommended further proceedings to consider designating AWS and Microsoft with Strategic Market Status, which would enable targeted remedies addressing lock‑in, egress fees, and licensing practices.
Taken together, these developments underscore that AI deal activity and hyperscaler market power are increasingly being analyzed as part of the same competitive landscape.
The Healthcare & Life Sciences and Financial & Insurance markets are among the most vulnerable in this space given their heavy reliance on hyperscalers to store and secure massive amounts of highly sensitive, confidential, private, and legally protected data. Click the links to see our related posts.
Why This Matters for Competition
Chokepoint deals are not inherently unlawful. But as framed by competition authorities and scholars examining AI markets, they warrant careful scrutiny because they can quietly reshape market structure. The familiar questions remain central:
- Do rivals retain meaningful access to essential inputs?
- Can systems interoperate, or does integration create lock‑in?
- Can customers realistically switch or multi‑home (use more than one platform or provider)?
Should We Leave the AI-Dominance Genie in the Bottle? Or Has the Cork Been Popped?
As antitrust scholars Maurice Stucke of the University of Tennessee and Ariel Ezrachi of University of Oxford have stressed in their article Antitrust & AI Supply Chains in Theoretical Inquiries in Law, “it is far easier to keep markets contestable than to try to restore contestability.” They add that competition agencies “missed or underappreciated” the dynamics that produced “a winner-take-most-or-all” digital economy, and warn that “feeble or delayed action” can allow emerging markets to tip before enforcement can meaningfully respond. That caution fits the deal pattern now playing out across the AI stack: once key inputs and distribution layers become structurally entrenched, the hard work is no longer detecting concentration—it is trying to unwind dependency after the fact.
Economists Susan Athey of Yale and Fiona Scott Morton of Stanford have similarly warned that the competitive stakes of AI infrastructure extend well beyond model performance or short‑term efficiencies. In their joint essay Cheap Goods for Everyone? The Impact of Market Power in Artificial Intelligence on Welfare and Inequality (Digitalist Papers, December 2025), they observe that “competition drives innovation, quality, and lower prices,” but that “without it, firms retain cost savings as profit, charge high prices, and stifle downstream innovation.”
Fiona Scott Morton has further cautioned that if AI is captured by a small number of dominant firms, “falling wages could coincide with persistently high prices,” a warning she reiterated in a January 2026 Yale Insights interview discussing the same work. Together, their analysis underscores why control over AI infrastructure and distribution layers—rather than headline model capabilities alone—may determine whether AI markets remain open and contestable, or tip toward durable concentration.
Conclusion
As reflected in recent scholarship, regulatory commentary, and the transactions catalogued in the Mogin Law AI Deal Table, the next phase of AI competition may be shaped less by who builds the best model, and more by who controls the stack’s chokepoints—compute, security and observability, data platforms, and enterprise distribution. These are the deals that can determine market structure long before traditional market shares are measured, and they are the transactions antitrust observers should be watching most closely.
Sources referenced include academic work by Susan Athey and Fiona Scott Morton on AI market power, research by Maurice Stucke and Ariel Ezrachi on AI supply chains, OECD competition analyses of AI infrastructure, and public statements by Google, Cisco, Palo Alto Networks, Workday, Accenture, and Databricks in connection with the transactions discussed.
Edited by Tom Hagy, Editor-in-Chief, Mogin Law Blog. Analysis of the Mogin Law AI Deal Table was performed with the assistance of AI, specifically Microsoft Copilot. This and all of are posts are intended for informational purposes only.
Contact Mogin Law
If you have questions about these or other competition law issues, contact us at Info@MoginLawLLP.com. Members of the press may write to us at Media@MoginLawLLP.com.