Book a Demo with LinqAlpha for More

Schedule

Key Themes

At the Morgan Stanley TMT conference, companies highlighted a structural shift where acute physical infrastructure constraints are driving pricing power, while software vendors aggressively pivot to "agentic" revenue models that monetize outcomes rather than user seats. To defend against disruption, incumbents are leveraging proprietary data, regulatory complexity, and liability frameworks as critical moats that generic AI models cannot replicate.

Below are are the key themes discussed across companies:

• Physical Scarcity & Infrastructure Yield: The AI supply chain has hit a wall of acute physical constraints, creating a landscape where NVIDIA declares "compute equals revenues" and Dell Technologies races to ship $1 billion in AI servers weekly against a massive $43 billion backlog. This insatiable demand is mirrored by CoreWeave, which reports 2026 is "broadly sold out" supported by a $67 billion backlog, while Lumentum remains sold out of critical Indium Phosphide lasers through 2027/2028. The strain on components is universal; Intel highlights severe Q1 shortages across wafers, substrates, and memory despite a mid-20s% rebound in server CPU demand, a tightness echoed by AMD across both GPUs and CPUs and HP, which is forcing price hikes to combat a sequential doubling of memory costs. As infrastructure expands to the edge, Akamai is navigating near-term margin compression (guided down to 23-26%) to fulfill a $200 million GPU contract-expanding from 20 to potentially 40 locations-similar to AT&T’s validation of physical density through its acquisition of Lumen assets. Meanwhile, Fortinet leverages proprietary ASICs to insulate margins from these component spikes, while Box positions itself not as a competitor to this compute intensity, but as the necessary "file system" for the resulting 100x proliferation of AI agents.
• The "Agentic" Pivot: Monetizing Outcomes Over Seats: Software vendors are aggressively shifting pricing from user seats to consumption or credit-based models to monetize the explosion of AI agents. Box argues that with agents potentially outnumbering humans 100-to-1, monetization must pivot to a "consumption volume-based activity model" treating agents as billable "machine users." This echoes Pegasystems, which abandoned user-based pricing years ago in favor of "cases" to ensure revenue grows even as automation reduces human headcount, a philosophy CrowdStrike is operationalizing via its mandatory "Flex" licensing to remove procurement friction. The industry is rapidly standardizing on this credit-based approach: ServiceNow reports a 55x surge in consumption of its "Assist packs," while Salesforce and Asana are similarly shifting toward "Agentic Work Units" and hybrid credit models to capture AI-driven consumption.
• The Liability Moat & Deterministic Accuracy: Incumbents are successfully arguing that "probabilistic" LLMs cannot handle "deterministic" high-stakes workflows without proprietary data and liability protection. Pegasystems anchors this defense, distinguishing its "deterministic workflow" for regulated industries where there is "zero tolerance for errors"-a precision generic AI cannot guarantee. This mirrors Vertex, which notes that 70% of tax rules are proprietary and offline, and Intuit, which leverages its "AI + HI" strategy to provide the liability shield customers demand. In the security realm, CrowdStrike warns that while AI excels at reasoning, cybersecurity demands absolute accuracy where "you don't get a second chance," a sentiment echoed by Box, arguing enterprises will never "vibe code" mission-critical governance infrastructure due to existential risk. Similarly, ServiceNow contends that generic models can only diagnose, whereas its integration allows for "workflow action," while Netflix and Uber maintain that human and physical realities remain the ultimate gatekeepers.

Day 1 (Mar 2, 2026)

Zscaler (ZS)

Zscaler management highlighted strong fiscal Q2 momentum, reporting 25% ARR growth and record large-deal volume driven by the Z-Flex flexible buying program ($290 million TCV). Key themes included the strategic pivot to AI security, with CEO Jay Chaudhry detailing the "AI Protect" launch and new solutions for securing AI agents. The company emphasized platform diversification, noting non-user-based revenue now exceeds 25% of new bookings. Management also expressed confidence in the Red Canary integration for AI SecOps and dismissed competitive concerns in the large enterprise market, reaffirming their $10 billion ARR goal.