Cisco’s AI Networking Revenue Has Crossed $5 Billion and Enterprise Data Centers Are Being Rebuilt for East-West GPU Traffic

Cisco disclosed in its Q3 FY2026 earnings call on May 14, 2026, that AI-related product orders had crossed $5 billion in the trailing twelve months — the first time Cisco has broken out AI networking as a separate revenue metric, reflecting both the size of the segment and the need to explain why Cisco’s networking hardware business is recovering after five consecutive quarters of enterprise spending contraction that followed the COVID-era overbuild cycle. Cisco’s Q3 FY2026 investor materials identify two distinct AI networking revenue streams: Cisco Nexus 9000 series switches being configured as AI cluster fabrics (replacing the traditional InfiniBand networking used in early GPU clusters with Ethernet-based connectivity that integrates with enterprise customers’ existing Cisco networking infrastructure), and Cisco Nexus HyperFabric, an AI-specific networking product launched in 2024 that provides a pre-configured fabric architecture optimised for the east-west GPU-to-GPU communication patterns that large language model training and inference require. The east-west traffic pattern is the defining architectural difference between AI data centers and traditional enterprise data centers: conventional enterprise networking was designed around north-south traffic — data moving between end-user devices and servers, or between on-premise infrastructure and the internet — where a hierarchy of distribution and access layer switches routes traffic through a central spine. AI training clusters require a fundamentally different architecture because the dominant traffic pattern is between GPUs within the same cluster during distributed training, where each GPU must communicate with dozens or hundreds of other GPUs simultaneously to synchronise gradient updates, parameter values, and activation states across the model training run. This east-west traffic pattern generates aggregate bandwidth demands of 400 to 800 gigabits per second per GPU node — orders of magnitude higher than the 10 to 25 gigabits per second per server that traditional enterprise networking was designed to support — requiring fabric architectures with near-zero latency, extremely high bandwidth-to-switch-port density, and lossless transport that preserves packet ordering across thousands of simultaneous flows. ARM Holdings’ compute subsystem royalties flow in part from the AI chip designs that generate this extreme east-west networking demand — every GPU sold into an AI training cluster creates a corresponding networking infrastructure requirement that Cisco’s HyperFabric products are designed to address.

Cisco’s competitive position in AI networking faces a structural challenge from Nvidia, which has its own high-performance networking division (formerly Mellanox) that sells InfiniBand interconnects — the dominant networking technology in GPU clusters before Ethernet became a viable alternative for AI workloads. InfiniBand’s historical advantage was its remote direct memory access capability, which allows GPUs to read and write each other’s memory without CPU intermediation, reducing the latency of gradient synchronisation during training by 2 to 3 times compared to standard Ethernet. Cisco’s Nexus HyperFabric and the broader Ultra Ethernet Consortium standard (of which Cisco is a founding member alongside AMD, Broadcom, and Intel) are attacking the InfiniBand dominance by demonstrating that modern 400G and 800G Ethernet fabrics with RoCE (RDMA over Converged Ethernet) achieve latency performance that is within 15 to 20 percent of InfiniBand in large-cluster training environments — a gap that Cisco argues is more than compensated by the operational advantage of running AI cluster networking on the same Ethernet infrastructure that enterprise customers already manage with Cisco tools, eliminating the need for a separate InfiniBand management layer that requires specialised expertise. The enterprise customer preference for single-vendor networking management is Cisco’s primary commercial advantage in AI networking: the 85 percent of Fortune 500 companies that run Cisco as their primary enterprise networking vendor have a strong default preference for extending that infrastructure into their AI cluster buildouts rather than introducing a new networking vendor and a new operational framework for AI-specific infrastructure. Cloudflare’s AI Gateway and edge inference products operate at the software layer above the physical networking fabric that Cisco provides — both companies are capturing value from the AI infrastructure buildout at different layers of the stack, with Cisco owning the physical transport layer and Cloudflare owning the API management and edge delivery layer above it.

What Cisco AI Defense Adds to the Networking Business

Cisco launched AI Defense in Q1 2026 as a security product specifically designed for enterprises deploying AI applications — addressing the security risks that emerge when employees and developers connect enterprise data to AI APIs (OpenAI, Anthropic, Google) without the visibility, access control, and data loss prevention mechanisms that IT security teams apply to conventional application traffic. AI Defense monitors and enforces policy on AI API calls from within the enterprise network perimeter: when a developer in a finance department submits customer account data to ChatGPT for analysis through an approved productivity tool, AI Defense classifies the data type, applies the enterprise’s data classification policy (marking customer PII as restricted and blocking the API call if the destination model provider’s data handling terms do not meet the enterprise’s compliance requirements), and logs the interaction for audit purposes. This is the same data loss prevention (DLP) function that Cisco’s existing security portfolio applies to email, USB transfers, and web uploads — extended to AI API traffic, which has emerged as the fastest-growing uncategored egress channel in enterprise networks since the commercial deployment of AI productivity tools accelerated in 2024 and 2025. Cisco’s integration of AI Defense into its existing security portfolio means enterprise customers can enforce AI API traffic policies through the same management console they use for all other network security policies, rather than deploying a standalone AI security tool from a new vendor. Palantir’s AIP platform addresses a complementary problem — ensuring that the AI-generated decisions and analytics that enterprises act on are grounded in verified enterprise data rather than model hallucinations — but the governance problem Palantir solves is at the application and decision layer, while Cisco AI Defense solves it at the network transport layer. Gartner’s networking and AI infrastructure research for 2026 projects that AI networking infrastructure — combining AI cluster fabrics, AI security tooling, and AI traffic management — will represent 35 percent of total enterprise networking spend by 2028, up from less than 10 percent in 2024, a trajectory that validates Cisco’s decision to break out AI networking as a separate revenue disclosure and to restructure its product development priorities around the AI data center buildout cycle.

Why the AI Networking Market Allows Cisco to Escape the Hardware Commoditisation Cycle

Cisco’s historical vulnerability in networking hardware has been commoditisation: white-box switching vendors (Arista’s whitebox alternatives, barefoot-based Broadcom-chipset switches programmed with open-source P4) have eroded Cisco’s pricing power in commodity 10G and 25G access-layer switching by offering comparable packet forwarding at significantly lower price per port. AI networking infrastructure is structurally resistant to this commoditisation pressure for two reasons specific to the AI cluster deployment context. First, AI cluster networking performance is directly tied to training throughput — a 20 percent improvement in fabric latency translates to a proportional improvement in training speed for distributed models, which at the scale of a 200,000 GPU cluster like xAI’s Colossus represents hundreds of millions of dollars in compute cost per training run saved or lost depending on fabric quality. Enterprises and hyperscalers buying AI networking infrastructure are willing to pay a meaningful premium for performance and reliability because the cost of a fabric-induced slowdown during a large training run far exceeds the cost difference between a premium and commodity switch. Second, AI cluster networking requires deep integration with GPU vendor drivers, RDMA network libraries, and cluster management software in ways that commodity white-box switches managed by generic open-source software cannot currently support with the same operational reliability as Cisco’s validated HyperFabric stack. Workday’s enterprise software business demonstrates a parallel commoditisation-resistance dynamic: HCM functionality in isolation is available from lower-cost vendors, but Workday’s data moat (1.5 billion skill inferences) and validated compliance workflows justify premium pricing for enterprise HR automation because the cost of errors in payroll, compliance, and headcount planning exceeds the cost of the software. Cisco’s AI networking premium is analogously justified by the training throughput cost of fabric underperformance at scale. The Wall Street Journal’s enterprise technology coverage through Q2 2026 frames Cisco’s AI networking pivot as the most important product strategy shift at the company since its 2015 to 2019 pivot to subscription software — a pivot that reduced Cisco’s hardware revenue dependence but took five years to reflect in financial results, while the AI networking cycle is producing immediate hardware revenue growth in the current quarter rather than requiring a multi-year transition period.

Workday Added AI Agents to Its HCM Platform and Enterprise HR Technology Has Entered Its Automation Phase

Workday reported $2.25 billion in Q1 FY2027 revenue (the quarter ending April 2026), a 16 percent year-over-year increase driven by subscription revenue growth in its Human Capital Management and Financial Management cloud products, while simultaneously rolling out its Illuminate AI product layer — which embeds AI agents directly into HR and finance workflows for headcount planning, skills gap analysis, pay equity audits, and dynamic organizational design — to its base of approximately 10,500 enterprise customers. Workday’s investor relations filings for Q1 FY2027 describe Illuminate as the company’s primary product investment priority for FY2027, with Workday allocating over 20 percent of its engineering headcount to AI feature development and targeting full Illuminate capability availability across its core HR and Finance product lines by Q3 FY2027. The commercial significance of Workday’s AI investment is not that it adds AI features to an existing product — every major enterprise software platform has announced AI integrations since 2023 — but that Workday’s HCM platform contains decades of structured organizational data (headcount histories, compensation records, performance ratings, skills inventories, org charts) that serves as the training and context foundation for AI models that are significantly more accurate for workforce-specific tasks than general-purpose LLMs prompted with the same data through an API. A Workday customer asking an AI system to model the headcount impact of a 10 percent revenue target increase can receive a scenario that draws on the organization’s actual role distribution, skill availability, historical headcount change patterns, and compensation benchmarks already stored in Workday — rather than a generic AI response that requires manual contextualization. Enterprise AI deployments at the scale of KPMG’s 276,000-seat implementation demonstrate that the organizations seeing the highest AI productivity returns are those where AI systems have access to structured organizational data — the kind of longitudinal, entity-linked data that Workday’s HCM platform accumulates over years of customer use — rather than those using general-purpose AI assistants over unstructured document repositories.

Workday’s AI differentiation in the HCM market rests on its Skills Cloud — a machine-learning system that maps an organization’s skills inventory by inferring from job titles, role histories, completed projects, certifications, and learning activity which skills each employee has demonstrated or developed, without requiring employees to manually self-report skills data. The Skills Cloud has been in production since 2020, and by 2026 it covers approximately 1.5 billion skill inferences across Workday’s customer base — a dataset that makes Workday’s workforce intelligence products qualitatively different from those of competitors that are building AI features on top of manually-maintained skills records. The Skills Cloud’s practical applications in the Illuminate product layer include internal mobility matching (identifying employees who have the skills needed for an open role without requiring a job application submission), pay equity analysis (identifying compensation gaps between employees with equivalent skill profiles in equivalent roles), and dynamic workforce planning (generating headcount scenarios based on skills supply and demand rather than static role counts). SAP SuccessFactors, Oracle HCM, and ADP each offer competing AI features in their HCM platforms, but none have a longitudinal skills inference dataset comparable in depth to Workday’s, because Workday’s platform has been ingesting structured HR events — role changes, promotions, project assignments, learning completions — from enterprise customers since 2012 and has a compound data accumulation advantage over competitors that built AI layers onto systems designed for data entry rather than continuous organizational intelligence. Salesforce’s Agentforce AI agent deployment in CRM demonstrates the pattern Workday is following in HCM: embedding AI agents that can execute multi-step workflows (schedule interviews, generate offer letters, update org charts) rather than just generate text responses, which is the operational automation tier that separates AI features that add to employee workload from AI features that reduce it.

What Workday’s Agentic HR Features Can Execute Without Human Approval

Workday’s Illuminate AI agent framework introduces a distinction that is commercially important for enterprise HR procurement: the division between AI-assisted workflows (where an AI generates a recommendation that a human reviews before action is taken) and AI-agentic workflows (where an AI executes a defined business rule without human review in the loop, except in cases flagged as exceptions). For routine, policy-constrained HR transactions — a leave of absence approval that meets the eligibility criteria defined in the company’s leave policy, a standard merit increase within the band defined for the employee’s job grade and performance rating, an onboarding task sequence completion notification — Illuminate’s agent mode can complete the transaction end-to-end without a manager or HR business partner reviewing the individual transaction. Workday’s enterprise customers define which workflows are eligible for agent-mode execution versus which require human-in-the-loop review, with Workday providing recommended exception criteria based on its cross-customer data on which transaction types generate reversal requests (an indicator that the human review step adds meaningful value) versus which almost never generate reversals (an indicator that the AI decision is consistently aligned with human judgment and the review step is pure overhead). Big tech’s workforce restructuring to fund AI investment has created a specific demand signal for Workday’s agentic HR capabilities: companies reducing their HR business partner headcount while growing their employee base need HR administration that can scale without proportional headcount growth, and AI agent automation of routine transactions is the mechanism that makes that ratio change operationally feasible. Gartner projects that by 2027, 30 percent of enterprise HR transactions that currently require manual processing will be fully automated by AI agent systems operating within policy guardrails — a projection that Workday’s Illuminate architecture is designed to capture at the platform layer rather than cede to point solutions or system integrators building automations on top of existing HRIS data. Gartner’s Human Capital Management research coverage positions Workday as a Leader in the HCM suite Magic Quadrant for 2026 with the highest score on completeness of vision, reflecting its AI integration roadmap and Skills Cloud data advantage over competitors whose AI features are add-ons rather than architecturally integrated with core data models.

Why Workday’s Competitive Position Depends on the Data Moat Holding

The strategic risk to Workday’s AI investment is not that SAP SuccessFactors, Oracle HCM, or Rippling will build better AI features in the next 12 months — it is that the general-purpose AI infrastructure (foundation models accessible via API, plus enterprise data integration tools like Snowflake or Databricks that can expose HR data to any LLM) could allow a new entrant to offer comparable AI workflow automation without Workday’s decade of structured HR data accumulation. Rippling, the fastest-growing HCM competitor in the US mid-market segment, has explicitly positioned its product architecture as a data integration layer that can connect any AI model to any HR data system — a strategy that bets the workforce intelligence use case can be solved at the integration layer rather than requiring Workday’s native data structure. Rippling’s approach works for organizations willing to invest in configuring the integration layer; Workday’s advantage is that its data model is already structured for workforce intelligence without integration work, making time-to-value for AI features shorter for enterprises that already have Workday as their system of record. OpenAI’s enterprise deployment consulting model represents an alternative AI delivery mechanism — where a consulting and integration layer translates general-purpose AI model capability into enterprise workflow automation — that competes with Workday’s native AI features for the same enterprise budget, but at a higher implementation cost and with less native integration into the transactional HR data that Workday already manages. Workday’s $2.25 billion quarterly revenue run rate, its 95 percent subscription revenue gross retention, and its 10,500 enterprise customer base give it the financial stability to sustain its AI infrastructure investment through a multi-year product transition — an investment cycle that pure-play AI application startups in the HCM space cannot match at comparable scale. The Wall Street Journal’s enterprise technology coverage through Q2 2026 characterizes Workday’s Illuminate rollout as the HCM market’s clearest example of incumbent enterprise software platforms using their proprietary data assets to resist AI-native startup disruption — a defense that is more durable than feature parity alone because it requires a competitor to replicate not just Workday’s technology but also the multi-year data accumulation that enterprise customers have contributed to Workday’s platform through their normal HR operations.

Palantir Crossed $1 Billion in Quarterly Revenue and AIP Has Become the Enterprise AI Decision Layer

Palantir Technologies reported $1.1 billion in Q2 2026 total revenue — the company’s first quarter above $1 billion and a 35 percent year-over-year increase that reflected accelerating commercial adoption of its Artificial Intelligence Platform (AIP) across US enterprise customers in manufacturing, healthcare, energy, and financial services verticals. Palantir’s investor relations disclosures for Q2 2026 show US commercial revenue reaching $490 million for the quarter (up 55 percent year-over-year), with US commercial customer count increasing to 465 from 262 in the same period one year prior — a growth trajectory driven by the AIP Bootcamp deployment methodology that Palantir introduced in mid-2023 as a structural change to how enterprises evaluate and adopt the platform. The AIP Bootcamp format — a five-day intensive engagement in which a Palantir team works with a client’s operational staff to build working AI-powered workflows on live production data within the client’s existing systems — compresses the enterprise software sales and proof-of-concept cycle from the 12 to 24 months typical for complex enterprise platform adoption to a single week that produces demonstrable operational output. The bootcamp model has proven particularly effective in manufacturing and industrial operations, where the gap between the data a company generates and the decisions it can act on with that data is large enough that a week of AIP workflow construction produces measurable throughput or cost improvements that justify multi-year platform contracts. Palantir’s government revenue segment — historically the company’s revenue base, covering US Department of Defense, intelligence community, and allied government contracts — reached $610 million in Q2 2026, growing more slowly (15 percent year-over-year) as the commercial segment has expanded to represent a larger share of total revenue.

What makes AIP commercially distinctive in the enterprise AI software market is the ontology layer — Palantir’s proprietary data modeling system that maps an organization’s operational entities (assets, personnel, workflows, decisions) into a structured data graph that AI systems can query and act on without requiring the client to restructure its underlying data infrastructure. Every enterprise that attempts to deploy AI on operational workflows faces the same foundational problem: the data relevant to a decision is scattered across multiple systems (ERP, CRM, MES, IoT sensors, logistics platforms) that were not designed to be queried together, and building a unified data layer is typically a multi-year data engineering project that precedes any AI deployment. Palantir’s ontology layer solves this by creating a semantic representation of the enterprise’s operations on top of existing systems without requiring data migration — the ontology maps where each piece of operational data lives and what it means in business terms, allowing AIP’s workflow tools to compose queries and actions across systems that have never interoperability. Enterprise AI deployments at the scale of KPMG’s 276,000-seat implementation demonstrate the range of approaches enterprises are taking to AI integration — from API-level model access at scale to platform-level operational workflow embedding — with Palantir’s approach sitting at the more deeply integrated end of the spectrum, where the AI system has direct access to operational data and decision workflows rather than acting as a text generation assistant layered over existing processes. The depth of integration that Palantir’s ontology enables is also the source of its sales cycle complexity: clients who adopt AIP are effectively committing to Palantir’s data modeling approach as the operational data layer for their business, a decision that requires more evaluation time than a seat-license productivity tool but produces a harder-to-displace position once adopted.

How AIP Bootcamp Changed the Enterprise AI Software Sales Model

The AIP Bootcamp format inverted the conventional enterprise software sales motion — which typically involves a multi-month request for proposal process, a structured proof of concept on synthetic or historical data, and a contract negotiation before any operational value is delivered — by front-loading the operational demonstration before the sales process concludes. In a standard bootcamp engagement, Palantir brings a team to the client site on day one with AIP already connected to the client’s production data systems (via connectors to SAP, Salesforce, Oracle, and major cloud platforms). By day three, operational staff who have never used Palantir’s tools are building AI-assisted decision workflows — inventory optimization routines, predictive maintenance triggers, logistics exception management — on live data from their actual operations. By day five, the workflow outputs are compared to historical baseline performance, and the quantifiable improvement becomes the basis for the contract discussion. The model works because it shifts the burden of proof from Palantir’s sales team to the client’s own operational data: the AI is not demonstrated on a curated demo environment but on the actual data the client works with, including the messiness, inconsistencies, and edge cases that enterprise data contains. OpenAI’s enterprise deployment company model represents a different approach to the same problem — building a consulting-adjacent service layer that handles enterprise integration complexity for clients who want to use frontier AI models but lack the in-house capacity to build operational integrations. The bootcamp model’s commercial success has prompted Microsoft Copilot Studio, ServiceNow, and other enterprise AI platform vendors to develop accelerated proof-of-concept formats that attempt to replicate Palantir’s compressed evaluation timeline, though without the proprietary ontology layer that makes AIP’s operational data integration distinctive.

What Palantir’s Commercial Growth Reveals About the Enterprise AI Decision Platform Market

Palantir’s Q2 2026 results reflect a broader shift in how enterprises are thinking about AI investment — moving from the productivity layer (AI assistants that help individual workers draft, summarize, and code faster) to the decision layer (AI systems that synthesize operational data and propose or execute decisions within business processes). The productivity layer market is dominated by Microsoft Copilot, Google Workspace AI, and Salesforce Einstein, all of which are distributed through existing enterprise software relationships and are measured in per-seat adoption rates. The decision layer market — where AIP competes — is measured in operational workflow coverage: what percentage of a company’s recurring decisions (which supplier to use, which maintenance task to prioritize, which shipment to reroute) are handled by AI-augmented systems rather than human-only judgment. Palantir’s commercial customer base skews toward industries where recurring operational decisions involve large amounts of structured sensor, logistics, or clinical data — manufacturing, energy, mining, healthcare — rather than the knowledge-work environments where productivity-layer AI excels. Big tech’s $725 billion AI infrastructure commitment is funding the model capability layer that both the productivity and decision tiers depend on, but Palantir’s commercial model captures value at the integration and workflow layer rather than the model layer — a position that insulates it from the commoditization pressure that is compressing margins at pure-play foundation model companies. Gartner’s analytics and AI platform research for 2026 positions Palantir as a Leader in its AI Decision Intelligence magic quadrant — a designation reflecting completeness of vision and execution ability in a market that Gartner defines as combining real-time operational data integration, AI-assisted decision workflow automation, and human-in-the-loop oversight infrastructure. Financial Times technology coverage through Q2 2026 frames Palantir’s commercial revenue inflection as evidence that the enterprise AI market is entering a second phase — beyond the initial experimentation period of 2023-2024, in which most enterprises ran pilots without committing to production deployment, toward a production deployment phase in which operational AI systems are being built and measured against hard performance metrics in the environments where a company’s actual revenue and cost structure live. The $1 billion quarterly revenue threshold positions Palantir as one of the few AI-first software companies that has converted the enterprise AI investment cycle into durable, contracted revenue at scale.

Salesforce Agentforce Is Generating Real Enterprise AI Revenue

Salesforce reported $9.8 billion in revenue for its fiscal Q1 2027 (ending April 2026) — up 8 percent year-over-year — with Agentforce, its AI agent platform for autonomous customer service, sales, and operations workflows, contributing to the acceleration of its Data Cloud and AI segment from a negligible revenue line to approximately $900 million in annualised recurring revenue. Salesforce’s Q1 FY2027 earnings disclosures show Agentforce-enabled deals accounting for a growing share of new business bookings — management stated that deals including Agentforce close at a higher average contract value than equivalent Salesforce platform deals without Agentforce, and that customer expansion rates on Agentforce accounts are running above the company’s historical expansion rate for similar customer cohorts. The commercial signal is the clearest validation Salesforce has produced for its AI platform bet since the original Agentforce announcement in September 2024.

Agentforce represents Salesforce’s answer to the question of where the enterprise CRM market goes after conventional software automation has been fully deployed. Salesforce’s core products — Sales Cloud, Service Cloud, Marketing Cloud — have been workflow automation platforms for two decades, helping companies manage customer relationships through structured processes and data capture. Agentforce extends that model into autonomous action: rather than automating a defined workflow where a human specified each step, Agentforce agents can interpret customer inquiries, pull relevant data from Salesforce’s Data Cloud, take actions (send emails, update records, create cases, schedule meetings), and escalate to human agents when the situation requires judgment beyond the agent’s configured scope. The distinction between conventional CRM automation and AI agent automation is the difference between a pre-programmed playbook and an agent that reads the situation and determines the appropriate next step. Multi-agent enterprise orchestration across the broader enterprise AI market has established that agentic AI workflows require orchestration infrastructure — Salesforce’s advantage is that it built its orchestration layer on top of the CRM data where most enterprise customer-interaction records already live.

What Agentforce Does in the Customer Service Layer

Agentforce’s highest adoption to date is in customer service — the business function where the volume of routine inquiries is highest and the cost of human agent time is most measurable. A customer service agent handling billing inquiries, order status questions, subscription changes, and account updates spends the majority of their working hours on queries that follow predictable patterns with well-defined resolution paths. Agentforce handles those queries autonomously — reading the customer’s history in Salesforce Service Cloud, identifying the appropriate resolution, executing the resolution (issuing a refund, changing an address, extending a subscription), and closing the case — without human involvement. The autonomous resolution rate that Salesforce customers are reporting for Agentforce-handled service volumes ranges from 40 to 70 percent depending on the complexity distribution of the query type, with human escalation handling the remainder.

The commercial case for customer service AI agents is the most straightforward in enterprise AI: the cost of a human agent handling a routine inquiry is typically $8-15 per interaction; the cost of an AI agent handling the same inquiry on Salesforce’s platform is approximately $0.50-2.00 depending on data retrieval and model call volume. An enterprise running 500,000 monthly service interactions that shifts 50 percent to AI agent handling reduces its service cost by $2-4 million per month while maintaining resolution quality for the inquiry types within the agent’s autonomous capability. Those economics are generating purchase decisions that do not require complex ROI modelling: the payback period is short enough that procurement teams can approve Agentforce without extensive internal analysis. Enterprise AI deployment at scale in professional services has demonstrated the same cost-displacement economics in knowledge work — Agentforce is producing the same dynamic in customer-facing service operations. Gartner’s AI customer service research projects that AI agents will handle 70 percent of routine enterprise customer service interactions by 2027, with Salesforce, ServiceNow, and Microsoft positioned as the primary platform vendors to capture that shift.

How Agentforce Competes With Microsoft Copilot

Microsoft’s Copilot for Dynamics 365 — its AI agent layer for enterprise CRM and ERP — is Agentforce’s most direct competitive threat in the enterprise market. The two products target the same enterprise buyer: companies managing large customer-facing teams who want AI to handle routine interactions and augment human agents on complex ones. The differentiation between the two is primarily in ecosystem affinity: companies already running Salesforce Sales Cloud, Service Cloud, and Marketing Cloud have a lower integration cost for Agentforce than for switching to Dynamics 365 and Copilot; companies already running Microsoft 365 across their organisation have a lower total-cost-of-ownership argument for Copilot given the licensing bundle advantages Microsoft offers. Enterprise CRM decisions in 2026 are consequently less about which AI agent product is superior in isolation and more about which CRM ecosystem the organisation is already committed to.

Salesforce’s response to the Microsoft bundling threat has been to expand Agentforce’s interoperability — announcing integrations with Slack (already a Salesforce property), Google Workspace, and Microsoft Teams — and to emphasise Data Cloud’s role as the source-of-truth data layer that makes Agentforce agents knowledgeable about the customer. The argument is that Salesforce holds more customer data in more enterprises globally than Microsoft Dynamics does, and that AI agents operating from more complete customer context produce better outcomes than agents with partial data access. Whether that data breadth advantage translates to measurable agent quality differences in production deployments is a question that enterprise buyers are evaluating through proof-of-concept projects in 2026. OpenAI’s enterprise deployment consulting arm has partnered with Salesforce customers on Agentforce implementations, which reflects the broader pattern of AI platform vendors partnering with model providers rather than building proprietary models — Agentforce agents run on multiple foundation models including OpenAI’s GPT series and Anthropic’s Claude depending on the task type and customer preference. TechCrunch’s Salesforce coverage through Q2 2026 documents the Agentforce customer base expanding beyond Salesforce’s traditional mid-market into Fortune 500 enterprise accounts where per-seat contract values are substantially higher.

The Sales Cloud AI Layer and What It Adds to the Product

Beyond customer service, Salesforce has deployed Agentforce capabilities into its Sales Cloud product — the CRM that manages pipeline, opportunity tracking, and account management for B2B sales organisations. Agentforce in the sales context operates as a sales coaching and next-best-action layer: analysing deal history, email correspondence, meeting notes, and competitive intelligence in Data Cloud to recommend specific follow-up actions for each opportunity in the pipeline. The product does not close deals autonomously — the judgment and relationship management that enterprise B2B sales requires remains human — but it surfaces the data patterns that experienced sales managers would identify manually, faster and more consistently than any human manager can across a large sales team.

The commercial uptake of AI in the sales workflow has been slower than in customer service because the ROI is less directly measurable. Customer service automation has a clear cost-per-interaction metric that allows ROI calculation without ambiguity. Sales productivity is more multivariable: whether an AI recommendation contributed to a deal closing is difficult to isolate from the many other factors that affect B2B sales outcomes. Salesforce addresses this measurement challenge by tracking win rate and deal velocity changes between Agentforce-assisted and non-assisted pipeline cohorts within the same customer organisation — a controlled comparison that has shown statistically significant improvements in the accounts Salesforce has published as case studies. The measurement approach is credible but curated: the published case studies represent customer organisations with strong data hygiene and well-configured Salesforce implementations, where the AI’s recommendations can draw on complete and reliable customer history. In organisations with fragmented data and inconsistent CRM adoption, the AI recommendations are less reliable, which is why Salesforce’s enterprise Agentforce sales process includes a Data Cloud readiness assessment before Agentforce deployment commitments are made.

Broadcom’s Custom AI Chips Power Google, Meta, and ByteDance’s Models

Broadcom reported AI revenue of $4.1 billion in its fiscal Q2 2026 — an annualised run rate above $16 billion — generated almost entirely from two sources: custom AI accelerator chips (XPUs) designed for specific hyperscaler customers, and the networking silicon that connects tens of thousands of those chips inside AI data centres. Broadcom’s Q2 FY2026 investor materials confirmed that Google, Meta, and a third unnamed hyperscaler (widely identified as ByteDance based on prior reporting) represent the majority of its AI XPU revenue, with each customer operating a multi-year design and production partnership that gives Broadcom the equivalent of a long-term contract in a market where competitors are typically evaluated project by project. The numbers position Broadcom as the second-largest beneficiary of AI infrastructure spending after Nvidia — a fact that receives significantly less attention than Nvidia’s market dominance because Broadcom’s AI chips are invisible to end users and absent from the public model benchmarking discourse.

The distinction between Broadcom’s XPUs and Nvidia’s GPUs is architectural and strategic. Nvidia’s H100, H200, and Blackwell series are general-purpose AI accelerators: programmable, flexible, capable of running any neural network architecture, optimised to perform well across training and inference for a wide range of model types. That generality is their value for AI research teams, startups, and enterprises that need a single hardware platform for varied workloads. The cost of generality is that general-purpose chips carry design overhead — memory bandwidth, programmability features, precision flexibility — that is unnecessary and expensive for a hyperscaler running a single well-defined workload at massive scale. Google’s TPU (Tensor Processing Unit) programme, which Broadcom has designed in close collaboration since TPU v4, starts from a different premise: what is the most efficient chip architecture for running Google’s specific matrix multiplication workloads at Google’s specific inference and training scales?

What Custom Silicon Actually Means for Google, Meta, and ByteDance

Google’s TPU v6 (Trillium), announced in mid-2025, delivers performance-per-watt improvements over the v5 generation that translate directly into the cost economics of serving Gemini inference at Google’s scale. Google processes hundreds of billions of AI-assisted queries monthly across Google Search AI Overviews, Gemini consumer, and Google Workspace features. At that volume, a 30 percent improvement in compute efficiency per FLOP compounds into billions of dollars of annual infrastructure cost reduction. The business case for the multi-year design investment in a custom chip is clear when the chip runs a single workload at that volume; the same case cannot be made for a company running diverse AI workloads in smaller quantities.

Meta’s MTIA (Meta Training and Inference Accelerator) chip family follows the same logic applied to Meta’s specific recommendation model workloads — the ranking and feed algorithms that process hundreds of billions of daily interactions across Facebook, Instagram, Threads, and WhatsApp. Meta’s recommendation workloads are among the highest-volume, most-stable inference tasks in existence: they run continuously, they are well-understood architecturally, and their compute requirements are predictable at a multi-year horizon. Custom silicon for a workload with those properties has a straightforward TCO argument. The MTIA programme represents Meta’s attempt to own the chip layer for its core revenue-generating models rather than remain dependent on Nvidia’s roadmap and pricing for that capacity. The Magnificent Seven’s $700 billion AI infrastructure commitment includes the custom silicon investment as a deliberate cost-reduction strategy embedded within total capex, not a separate line item.

How XPUs and Networking Drive Broadcom’s AI Revenue Mix

Broadcom’s AI revenue is roughly split between two product categories. The first is the XPU chip design and production business — Broadcom designs the chip in partnership with the hyperscaler, manufactures it at TSMC using N3 or N2 process nodes, and earns revenue on chip sales. The second, and in some quarters the larger contributor, is AI networking silicon: the Tomahawk and Jericho ethernet switch chips that interconnect the accelerator clusters inside AI data centres.

AI training requires tight coordination among thousands of accelerators running in parallel; the interconnect between them must move data at rates that keep the accelerators fed without creating bottlenecks. Broadcom’s 51.2 Tbps Tomahawk 5 ethernet switch is the dominant switching silicon for high-bandwidth AI cluster interconnects, with deployments at every major hyperscaler’s AI data centre construction programme. Nvidia’s Blackwell infrastructure uses a mix of NVLink (Nvidia’s proprietary interconnect) for dense in-rack coupling and ethernet (often Broadcom Tomahawk-based) for rack-to-rack fabric — meaning Broadcom’s networking business benefits from Nvidia deployments as well as from custom XPU deployments that do not use Nvidia at all. The networking revenue is effectively a toll on all AI data centre construction regardless of which accelerator chip is inside.

Why Nvidia Hasn’t Lost and Why That May Change

Nvidia’s dominance in AI compute is not threatened by Broadcom’s XPU business in the near term, and the reason is timing and scope. Custom silicon development takes 3-4 years from design inception to volume production; the workload must be stable and large enough to justify the design investment; and the chip must be maintained and iterated in partnership with a single customer who accepts the risk of the design not performing as expected. These conditions apply to a small number of hyperscalers with the largest and most stable AI workloads. For the 99 percent of AI compute buyers who are not at Google or Meta scale — enterprises, cloud customers, AI startups, research teams — Nvidia’s general-purpose GPUs with their mature software ecosystem (CUDA, cuDNN, TensorRT) remain the only viable option.

The long-term dynamic is that custom silicon’s share of total AI compute will grow as more hyperscaler-scale workloads mature and as the design ecosystem improves. Hyperscaler cloud capex is increasingly allocated toward custom silicon as a percentage of total chip spend, and Amazon’s Trainium3 (a Broadcom-adjacent programme) and Microsoft’s Maia 2 represent additional major hyperscalers moving down the same path. Whether Broadcom retains the dominant position in XPU design-and-manufacture or faces competition from other chip design firms as the market grows is the strategic question for the post-2026 period; for now, its three-customer concentration in XPUs and its networking silicon monopoly position give it an AI revenue trajectory that no other semiconductor company outside Nvidia can match.

ServiceNow Crossed $3.5 Billion Quarterly Revenue on AI Workflows

ServiceNow’s Q2 FY2026 results confirmed the company’s subscription revenue has crossed $3.5 billion in a single quarter — the first time any pure enterprise workflow platform has reached that milestone without a hardware or consumer business attached. ServiceNow’s Q2 FY2026 investor release reported subscription revenue of $3.52 billion, a 26 percent year-over-year increase, with the company’s AI-embedded SKUs now representing a material portion of net new annual contract value. The result positions ServiceNow as one of the five largest pure-software subscription businesses in the world by quarterly revenue, alongside Salesforce and Oracle — neither of which competes with it on the same workflow terrain.

The growth trajectory matters because it has occurred concurrently with a period when enterprise technology spending has bifurcated sharply. Capital investment in AI infrastructure — data centres, GPU clusters, foundation model training — has commanded the majority of headlines, while application-layer spending has faced tighter scrutiny. ServiceNow has outgrown that scrutiny because its platform delivers measurable process automation outcomes that enterprise finance teams can audit: ticket deflection rates, resolution time compression, headcount-to-workflow ratios. That auditability — the ability to show a cost centre leader what the platform is actually doing — separates ServiceNow from AI tools where the value proposition is diffuse and the token cost is real, as the enterprise AI cost reckoning increasingly documents.

Now Assist’s Commercial Traction Across Enterprise Accounts

ServiceNow’s AI layer — branded Now Assist — integrates generative AI capabilities directly into the workflows that enterprise teams already run on the Now Platform: IT service management, HR case handling, customer service operations, and IT operations (AIOps). The commercial adoption pattern differs from point AI tools because Now Assist does not require a separate procurement conversation or a new integration project. Enterprises already running ServiceNow activate Now Assist as an upgrade to existing workflows rather than buying a new product. That distribution advantage has produced accelerating AI SKU attach rates: more than 40 percent of ServiceNow’s new enterprise contracts in Q2 FY2026 included at least one Now Assist-tier product, compared with 18 percent in Q2 FY2025.

The practical deployment cases are narrower than general-purpose AI tools and more directly valuable for that reason. Now Assist for ITSM generates incident summaries and suggested resolutions at the time of ticket creation, reducing the mean time to resolution on tier-1 incidents by a measurable factor without requiring analyst review at the first stage. The AIOps module correlates event noise across monitoring systems before a human operator touches the alert queue — a function that becomes more valuable as infrastructure complexity grows. Enterprise-scale AI deployment programmes, which are now reaching into the hundreds of thousands of knowledge worker seats, require the kind of workflow-embedded AI that integrates into existing ticketing and service delivery systems rather than sitting adjacent to them. ServiceNow’s platform architecture is the delivery mechanism that general-purpose LLM APIs are not.

Microsoft Is Not ServiceNow’s Competitor in This Category

The most analytically important feature of ServiceNow’s market position is that Microsoft Copilot — despite its ubiquity in enterprise IT discussions — is not a direct substitute for the Now Platform in the ITSM, HR service delivery, or enterprise workflow automation categories. Microsoft Copilot integrates with Microsoft 365 applications and Azure DevOps. It does not natively manage the incident lifecycle, the change approval workflow, the service catalogue, or the configuration management database that ServiceNow’s ITSM module governs. An enterprise CIO using ServiceNow for IT operations and Microsoft 365 for productivity is buying distinct products for distinct functions. The overlap exists at the edges — AI-assisted search, automated email routing, natural language query — but not at the core process layer.

This structural separation is worth precision because the enterprise AI narrative has generated significant market-level anxiety about platform consolidation risk. The concern is that Microsoft, Google, or Salesforce will absorb the workflow management category through AI capability expansion the way productivity suites absorbed standalone document management in the 1990s. Microsoft’s own platform monetisation cycle shows the pressure that hyperscalers face from customer consolidation demands, but the ITSM category has resisted that pressure precisely because the switching costs of Now Platform migrations are high and the platform’s depth in process automation has not been replicated by any hyperscaler-native offering. ServiceNow’s Q2 ACV retention metric — net new ACV from existing customers minus ACV lost to churn — remained above 120 percent for the fourteenth consecutive quarter, which is the retention signal that the consolidation-risk thesis would require to decline first.

What $3.5 Billion in Subscription Revenue Tells Enterprise Buyers

At $3.5 billion quarterly subscription revenue, ServiceNow has reached the scale at which platform viability is no longer a meaningful procurement risk. Enterprise technology procurement teams have a multi-year investment horizon for platforms that govern mission-critical operations; they price the vendor risk of a platform failure or acquisition into their TCO calculations. The scale threshold below which procurement teams require acquisition or bankruptcy provisions in enterprise contracts is generally assessed at around $2 billion annual recurring revenue for vertical workflow platforms. ServiceNow has exceeded that threshold by more than 7x. The relevant risk question for CIOs and CPOs reviewing ServiceNow renewals in H2 2026 is not whether the platform will exist in five years — it will — but whether its AI capability roadmap justifies the premium pricing relative to legacy ITSM alternatives.

On that question, Gartner’s analysis of the ITSM and enterprise service management market has consistently placed ServiceNow in the strongest position for AI-augmented workflow automation, distinguishing between the generative AI feature parity that legacy vendors have achieved at the surface level and the architectural depth of integration with live operational data that ServiceNow’s platform provides at the process layer. The Q2 results — growing 26 percent at $3.5 billion in a period when enterprise technology spending broadly decelerated — confirm that the architecture distinction is converting into commercial outcomes for the platform’s customers and for the platform’s valuation, which has expanded from roughly 12x ARR at the start of 2025 to approximately 15x forward ARR at current trading levels.

The Boring-Software Thesis Behind ServiceNow’s AI Quarter

Paul Graham’s recurring observation about startups applies in inverted form to ServiceNow: the most defensible software businesses are usually the ones that sound boring at dinner parties. Ticket routing, change management, employee onboarding workflows — nobody ever raised a seed round on enthusiasm for those categories. But boring categories share a structural property that glamorous ones lack: the customer’s alternative to the product is not a competitor, it is institutional chaos. A company that rips out its workflow platform does not switch to a rival so much as it reverts to email threads and spreadsheet trackers. That asymmetry is the foundation under ServiceNow’s revenue durability, and it explains why AI monetisation is landing faster here than in most enterprise software.

The reason is mechanical rather than visionary. AI features sell when they attach to a workflow the customer already runs and already measures. ServiceNow’s installed base has spent a decade encoding its operational processes — approvals, escalations, fulfilment steps — into the platform. An AI layer that compresses any of those steps produces a measurable time saving against a baseline the customer already tracks. Compare that to the generic enterprise chatbot, where the buyer has to invent both the use case and the measurement before any value shows up. The boring company gets to skip the hardest part of AI adoption: proving that the work being automated was real work.

The risk in the thesis is the same one Graham flags for any company whose moat is accumulated configuration: the moat holds only while the cost of re-encoding those workflows elsewhere stays high. Agentic AI is precisely the technology that could collapse that cost — an agent that can observe and reconstruct a company’s approval chains from its communication exhaust would do to workflow platforms what data-migration tooling did to proprietary file formats. ServiceNow is betting it can build that agent layer itself before someone builds it against them. The Q2 numbers say the bet is working so far. They do not yet say anything about whether the moat survives the technology that is currently funding it.

Arm’s Server Market Share Is Accelerating Past Intel

AWS Graviton4, the fourth generation of Amazon’s Arm-based custom processor, now runs approximately 40% of all general-purpose compute instances on AWS — up from 28% two years earlier. The figures come from Amazon’s Graviton4 general availability announcement and represent the fastest rate of architectural share gain in the hyperscaler compute market. Microsoft’s Azure Cobalt 100 (Arm-based, launched commercially in late 2024) and Google’s Axion processor (Arm-based, in broad availability across GCP regions from Q1 2026) mean that all three major cloud providers now have in-production Arm silicon carrying material workload fractions.

Intel has held dominant data center CPU revenue for two decades. The server processor market is not going to zero for x86 — legacy workloads, Windows Server deployments, and specific latency-sensitive applications continue to favour Xeon — but the trajectory of new workload placement is running against Intel and toward Arm at a rate that cannot be explained by price alone.

Performance-Per-Watt: The Economics Driving Hyperscaler Choice

The hyperscaler adoption of Arm processors is principally an economics decision, not an architectural preference. AWS has published benchmark data for Graviton4 showing 40% better price-performance than comparable x86 instances for web serving and general-purpose application workloads. The efficiency advantage is larger in workloads that benefit from Graviton’s memory bandwidth architecture — data analytics, distributed computing frameworks, and containerised applications at scale.

Power consumption is the amplifying factor at hyperscaler scale. A 30% improvement in performance-per-watt translates directly to data center capacity density and energy cost reduction. Hyperscalers committed more than $700 billion in AI infrastructure capital in 2026, and power and cooling costs are a primary constraint on how much compute that capital can deliver. A data center architecture that extracts 30% more useful compute per megawatt of power capacity is worth substantially more than its benchmark headline suggests.

The AI training workload is not Arm’s primary battlefield — Nvidia’s GPU dominance in training is intact, and Nvidia’s $81.6 billion revenue quarter is evidence of that dominance compounding. But Arm is taking share in the inference and general-purpose compute layers that sit alongside the GPU clusters: the CPU instances that handle model orchestration, token routing, request preprocessing, and application logic around AI pipelines. This layer is large and growing.

Arm Holdings’ Royalty Model Is Changing with the Market

The economics of Arm’s success at the hyperscaler level are structurally different from Arm’s traditional licensing model. Arm Holdings generates revenue through technology licensing (upfront fees for architecture access) and royalties (per-unit fees on shipped chips). Traditionally, royalties came from the consumer electronics cycle — smartphone chips, embedded devices, microcontrollers. The hyperscaler custom silicon wave — AWS Graviton, Microsoft Cobalt, Google Axion, Ampere Computing — creates a royalty revenue stream from data center chips that did not exist at meaningful scale five years ago.

Arm Holdings’ FY2026 results showed infrastructure royalty revenue growing at approximately 60% year-on-year, driven by hyperscaler silicon shipments. The infrastructure segment is now large enough to be a material factor in Arm’s total royalty mix. The practical consequence for Arm’s business model is that its revenue is increasingly linked to data center chip shipments rather than smartphone shipments — a market that is growing faster and carries higher per-chip royalty values.

Intel’s response has been structurally constrained by its foundry problems. Producing server CPUs competitive on performance-per-watt requires manufacturing process nodes that Intel’s own fabs have struggled to deliver reliably at volume. The Intel 18A process node — Intel’s plan to reclaim process leadership from TSMC at the 18-angstrom node — has been in an extended qualification period. Cloud infrastructure spending patterns show hyperscalers continuing to expand Arm-based capacity while Intel’s equivalent design wins in the same tier have not materialised at expected volume.

Where x86 Remains Defensible

The scenario in which Arm displaces Intel entirely from server infrastructure is not the base case. Intel’s server CPU business retains defensible positions: Windows Server workloads, enterprise applications certified on x86 architecture, and workloads where instruction set architecture compatibility is a constraint rather than a performance optimisation. For organisations running decades of code compiled against x86, re-architecting for Arm is a project that competes with other priorities. The largest enterprise IT organisations are not going to recompile their entire application estate for Arm performance gains at their specific workload scale.

AMD’s EPYC processors have maintained their own gains in this market — AMD has taken genuine share from Intel in server CPUs and has done so on a more competitive process node. But AMD is running the same x86 architecture, which means AMD benefits or loses from Arm’s share gains in roughly the same proportion as Intel. The architectural competition is x86 against Arm, not Intel against AMD, in the market that matters: new workload placement at hyperscaler scale.

The rate of Arm’s share gain over the next two years will be determined primarily by how quickly the enterprise (non-hyperscaler) server market adopts Arm, which depends on software ecosystem maturity and ISA compatibility tooling rather than processor performance benchmarks. In the hyperscaler market, the architecture decision is already largely made. The question for 2027 and 2028 is whether the enterprise market follows the hyperscalers’ lead — or whether the software compatibility constraint keeps x86 dominant in that segment for another decade while Arm consolidates the cloud.

Arm’s Counter-Positioning and the Limits of Intel’s Response

Hamilton Helmer’s Power framework identifies Counter-Positioning as one of the most durable competitive advantages — and one of the most strategically awkward to defend against. A challenger adopts a superior business model that an incumbent cannot copy without severely damaging its existing business. Arm’s position in the server market is a near-textbook example. The superior performance-per-watt economics of custom Arm silicon — visible in AWS Graviton4, Ampere Altra, and Microsoft Cobalt — are achievable only by companies willing to absorb the multi-year investment in custom silicon design. Intel’s response requires doing exactly what would cannibalise its volume server CPU business before an alternative revenue source is ready.

The counter-positioning mechanism here is specific: Intel’s existing x86 server business is sustained by a software compatibility moat that is worth billions in annual revenue. Custom Arm silicon deployment at hyperscaler scale requires the hyperscalers to invest in ISA-level software porting and optimisation — a cost they absorb because the performance-per-watt payoff justifies it at their workload volumes. Intel defending its x86 position means resisting the move to custom silicon; Intel following the hyperscalers into custom silicon means acknowledging that x86’s performance-per-watt economics are inferior for cloud workloads and triggering a re-evaluation of the entire enterprise x86 installed base.

The Power framework also offers the concept of Switching Costs as a separate power type — and here the picture for Intel is more complex. The enterprise (non-hyperscaler) server market is insulated from Arm adoption by software compatibility switching costs that the hyperscalers have already absorbed but that a manufacturing company running ERP workloads on x86-native enterprise software cannot easily replicate. Intel’s remaining durable position is in this enterprise segment, where switching costs keep x86 relevant even after the hyperscaler market has largely moved to custom Arm. The strategic question for Intel is whether defending enterprise x86 yields enough value to justify the investment, or whether the margin compression from hyperscaler share loss makes the enterprise segment insufficient as a long-term foundation.

Arm’s IR commentary on hyperscaler royalty growth rates — up significantly year-on-year — reflects the beginning of the monetisation arc for a decade-long silicon design investment cycle. The Power at scale for Arm is not the ISA licensing model itself (easily copied in theory, if not in practice) but the ecosystem depth: the compiler toolchains, the cloud-native software stack, the silicon design expertise concentrated at the hyperscalers, and the benchmark performance record being built deployment by deployment. That ecosystem constitutes a genuine Process Power advantage that Intel is not positioned to replicate on a two-year timeline, regardless of how aggressively it invests in counter-architecture development.

Intel’s 18A Process Node: Whether the Company’s Foundry Reset Can Actually Threaten TSMC

Intel began risk production of its 18A process node in Q1 2026 — a milestone Intel’s leadership called “the most significant technical achievement in the company’s modern history.” 18A is Intel’s gate-all-around (GAA) transistor implementation, competing directly against TSMC’s N2 node on density and power efficiency. If 18A delivers on its specifications, Intel Foundry Services has a credible leading-edge logic product for the first time in a decade. If it does not, the foundry strategy Intel has staked approximately $45 billion in capital on over the past four years will face terminal questions from its investors and its customers.

The technical data published so far suggests Intel’s claims are partially supported and partially aspirational — which, at this stage of the risk production cycle, is better than the foundry strategy’s history since 2021 warrants.

18A Technical Specifications Against TSMC N2

Intel’s published 18A specifications claim approximately 10% performance improvement and 30% power reduction versus Intel 3 (its previous generation), at a transistor density comparable to TSMC N3E. Against TSMC’s N2 — which Intel is directly positioning 18A to compete with — the published claim is performance parity at comparable power, with Intel arguing a cost advantage from its RibbonFET (GAA) implementation and its integrated backside power delivery (PowerVia).

Independent foundry analysis from SemiAnalysis — the most technically rigorous public semiconductor analysis available — assessed 18A as capable of competing with TSMC N3E but not yet definitively at N2 parity. The integrated backside power delivery is genuinely novel and delivers meaningful power efficiency improvement; the RibbonFET implementation is technically comparable to TSMC’s GAA but is a first-generation production implementation that will require yield learning before it matches TSMC’s production maturity.

Yield is the operative variable. TSMC’s N2 has been in risk production since late 2025 with customer tape-outs; Intel’s 18A is in risk production now, approximately six months behind. At risk production, both nodes are operating below commercial yield — the percentage of functional chips per wafer that makes production economically viable. TSMC’s typical ramp from risk to commercial yield takes 12-18 months. Intel’s recent history (delays and yield problems on Intel 4 and Intel 3) makes the same timeline optimistic, but Intel’s manufacturing organisation has been substantially restructured under Pat Gelsinger and his successor, and the current 18A execution has proceeded more closely to schedule than its predecessors.

The IFS Customer Pipeline

Intel Foundry Services’ commercial viability depends on attracting customers who will commit multi-year wafer agreements at volumes that utilise Intel’s fab capacity. The current 18A customer pipeline includes Microsoft (confirmed via public disclosure), a US Department of Defense programme, and several undisclosed customers that Intel has characterised as “hyperscale and defence.”

Microsoft’s 18A commitment is the most commercially significant disclosed agreement. Microsoft has announced plans to use Intel 18A for undisclosed chip designs — likely custom AI accelerators for Azure rather than x86 consumer products — with wafer production scheduled to begin as 18A ramps to commercial yield in 2027. The Microsoft commitment represents a validation from a hyperscaler that has the engineering resources to evaluate foundry alternatives rigorously and the financial credibility to make the commitment meaningful.

The Apple relationship, which Intel and Apple have discussed publicly, remains uncertain. Intel’s Apple chip talks have centred on whether Apple would use Intel Foundry for future A-series or M-series silicon production alongside TSMC, providing geographic diversification for Apple’s most critical chip production. Apple has not committed publicly, and the timeline for any Apple IFS production would be 2028 at earliest given Apple’s multi-year chip design lead times. But a disclosed Apple commitment would be transformative for IFS’s commercial credibility in a way that even the Microsoft deal is not — Apple’s chip volumes are the single largest leading-edge logic customer in the world.

Intel’s Financial Position Under the Foundry Bet

Intel’s capital investment in its manufacturing turnaround has been the largest in US semiconductor history: approximately $20 billion in 2024 capital expenditure, $18 billion planned for 2025, and $14 billion in 2026 as the fab buildout matures and operating costs stabilise. The CHIPS Act provided approximately $8.5 billion in direct funding and approximately $11 billion in loan guarantees, reducing the net capital burden — but Intel is still running at negative free cash flow as the foundry investment scales ahead of revenue.

Intel’s Q1 2026 financial results showed IFS revenue of approximately $4.7 billion — up 8% year-over-year but still substantially below the $20 billion annual IFS revenue target that management has set for 2030. The gap between current IFS revenue and the target requires signing major external customers (currently, IFS revenue is dominated by Intel’s own product designs). At current external customer win rates, the 2030 target requires signing 3-4 major hyperscaler or fabless chip customer relationships within the next 18 months — a timeline that aligns with 18A reaching commercial yield and customer tape-out volumes.

What the Terafab Discussion Signals

The announced discussions about a potential Terafab programme — a large-scale US semiconductor manufacturing joint venture involving Intel, US government funding, and potentially industry partners including Elon Musk’s xAI — adds a geopolitical dimension to Intel’s foundry trajectory. The programme, which has not moved beyond discussion and MOU stages, would potentially provide additional capital for fab expansion at a scale that Intel could not fund independently.

The Terafab concept is driven by the same national security logic as the CHIPS Act: the US government wants leading-edge logic production capacity on American soil that is not dependent on TSMC’s Taiwan concentration. Intel is the only American company with the engineering capability to attempt this. Whether Terafab actually forms and at what terms are unknowns, but the discussion itself signals that Intel’s political capital with the US government remains intact — a non-trivial asset in the current semiconductor policy environment.

The Honest Assessment

Intel’s 18A represents the company’s most credible foundry technology since 2016. The GAA implementation is technically sound, the backside power delivery is genuinely innovative, and the execution to date has been closer to schedule than any of Intel’s prior leading-edge node programs since Cannon Lake. These are real improvements.

Against this is the context: TSMC’s N2 had its first 18 months of capacity pre-sold before production started, its CoWoS advanced packaging capacity is being doubled, and its gross margins are at 53%. TSMC’s existing customer relationships, production maturity, and supply chain ecosystem represent a structural moat that Intel cannot close through process node parity alone — it requires convincing customers to adopt Intel’s foundry infrastructure, which means qualification cycles, co-investment in packaging development, and management bandwidth commitments that customers will not make without a compelling risk-adjusted case.

The 2027-2028 window — when 18A reaches commercial yield, the first external customer chips begin production, and the Microsoft tape-out results become evaluable — will provide the definitive answer to whether Intel’s foundry reset produces a real competitor to TSMC or a perpetually-promising-but-second-tier alternative. The bet Intel’s investors have made is on the former. The semiconductor industry’s history of the past decade suggests caution about that bet. The 18A execution to date suggests the caution should be moderate, not absolute.

Google I/O 2026: Gemini 2.0 Ultra, Android 16, and the Search Reinvention That Puts Google’s Core Business at Risk

Google I/O 2026, held in mid-May at the Shoreline Amphitheatre, was simultaneously Google’s most impressive technical showcase in years and the clearest public statement yet of the company’s central tension: how to deploy the AI capabilities that could make Google Search obsolete without making Google Search obsolete.

The announcements — Gemini 2.0 Ultra, AI Overviews’ expansion, Android 16’s deep Gemini integration, Project Astra’s progress toward persistent multimodal AI, and the continued evolution of NotebookLM — were technically impressive across the board. But the strategic subtext beneath each announcement was the same: Google is trying to turn the threat of AI-disrupted search into a durable advantage before someone else does it to them.

Gemini 2.0 Ultra: The Benchmark Leader That Matters Less Than It Should

Gemini 2.0 Ultra debuted at I/O 2026 — building on the agentic shift previewed earlier in the I/O keynote — with benchmark scores that establish it as the leading publicly-available foundation model on several major evaluations. On the MMLU Pro reasoning benchmark, Gemini 2.0 Ultra scores 91.4 — above GPT-4.5’s 89.7 and Claude 3.7 Opus’s 90.1. On coding benchmarks including HumanEval and LiveCodeBench, Gemini 2.0 Ultra similarly leads the pack — and the Flash tier compression that has played out earlier in 2026 means the price-performance advantage extends down the model stack. On multimodal benchmarks, it holds a more commanding lead: Google’s investment in video and audio understanding, built on top of its YouTube training data advantage, produces measurable capability improvements on video comprehension tasks that text-focused models cannot match.

The benchmark victory is genuine. The commercial implication is more complicated.

Enterprise buyers increasingly understand that benchmark scores predict model capability on well-defined tasks but do not fully predict real-world deployment reliability, instruction-following consistency, or safety behaviour. The enterprise sales cycle for foundation model access runs through procurement teams that prioritise vendor stability, compliance documentation, and integration support over benchmark rankings. In this environment, being the benchmark leader is a marketing advantage, not a decisive commercial one.

Google’s distribution through Google Cloud’s Vertex AI platform is its more durable competitive advantage. Gemini 2.0 Ultra access through Vertex AI means enterprise buyers already on GCP — Google’s estimated 30% share of enterprise cloud deployments — can add Gemini access to their existing vendor relationship without new procurement processes. For Google, the benchmark win matters primarily as permission to be in the evaluation shortlist; the distribution advantage is what converts evaluations to contracts.

AI Overviews and the Search Revenue Question

The most consequential and most carefully managed announcement at I/O 2026 was the expansion of AI Overviews — Google’s AI-generated search summaries that appear above organic results. AI Overviews now trigger for approximately 40% of Google Search queries in the US, up from 25% at launch and the 10% in the experimental phase. The expansion includes more categories: shopping queries, local business queries, and multi-step research queries now routinely receive AI Overview summaries.

The commercial tension is explicit: when an AI Overview answers a user’s question directly in the search results page, that user has less reason to click through to a website. Fewer click-throughs mean fewer opportunities for Google’s cost-per-click advertising to generate revenue. AI Overviews that are monetised with ads embedded in the summary itself produce lower CPMs than traditional search ads (because the user is reading rather than actively seeking to transact). The revenue-per-query economics of AI-augmented search are structurally lower than the revenue-per-query economics of traditional search.

Google’s response to this tension has been to move fast and shape the market before anyone else can. If AI search summaries are inevitable — which Google’s own data suggests, given user satisfaction scores for AI Overview results — then it is better for Google to cannibalise its own click-through revenue than to allow a competitor to capture the AI search market and cannibalise Google’s entire revenue base.

The bet is that AI-augmented search, despite lower per-query revenue, increases total query volume and total user time in the Google ecosystem sufficiently to offset the per-query revenue decline. Early data from Google’s advertising team supports this: average queries per user per day increased approximately 18% in markets where AI Overviews have been fully deployed for more than six months. If per-query revenue falls 20% but queries grow 18%, the net revenue impact is manageable — and if query growth continues to compound while per-query revenue stabilises, the long-term economics improve.

Android 16: Gemini Everywhere

Android 16, previewed at I/O 2026 for release to Pixel devices in Q3 2026, ships with Gemini as the system-level AI — replacing Google Assistant throughout the operating system. The integration is materially deeper than previous Gemini rollouts: Gemini has access to all on-screen content, the device’s notification history, calendar, contacts, Gmail, and Google Photos, enabling the contextual awareness that Apple Intelligence’s Siri has been attempting to achieve.

The Android 16 Gemini integration is significant for two reasons beyond user experience. First, the scale: approximately 3 billion active Android devices will eventually run Gemini-integrated Android, giving Google a training signal and product feedback loop that no competitor can match. Second, the data advantage compounds over time — Gemini learning from billions of Android interactions (with appropriate privacy controls) builds a behavioural model of how people actually use AI-augmented mobile operating systems that will improve Gemini’s on-device performance in ways that are structurally difficult to replicate.

The competitive comparison to Apple Intelligence is inevitable and instructive. Apple’s on-device AI runs 3-7B parameter models; Google’s Pixel-native Gemini Nano (the on-device component) has been expanded to larger model sizes with the A19-class Tensor chip in Pixel 10. The on-device vs cloud-dependent architecture debate continues, but Android 16’s approach — a hybrid that runs common tasks on-device and escalates complex tasks to cloud Gemini — is more pragmatic than Apple’s privacy-first on-device purist position.

Project Astra: The Persistent Multimodal Assistant

Project Astra, Google DeepMind’s research project for a persistent, multimodal AI assistant, showed its most advanced capabilities at I/O 2026. The demonstration showed an AI that maintains persistent memory across conversations (remembering context from sessions days earlier), understands video in real time through a phone camera, and can navigate complex multi-step tasks by combining visual understanding, web access, and long-form reasoning.

Astra is not a shipping product — the full vision remains a research demonstration. But the components are real and progressively being deployed: Gemini Live (real-time voice conversation), camera-based contextual awareness in the Gemini app, and memory features that persist across conversation sessions. The I/O 2026 demonstration showed these components operating more fluidly than in any previous public demo, suggesting the gap between research vision and shipping product has narrowed.

The strategic importance of Project Astra is not its current state but what it signals about Google’s capability roadmap. If Astra’s full vision ships — a persistent AI that knows your history, understands your environment in real time, and can act autonomously on your behalf — it represents a shift from search as query-and-response to search as continuous ambient intelligence. Google’s position at the centre of that paradigm is more defensible than its position in a world of competing AI chatbots, because the data infrastructure required to make Astra work at scale is something only Google (with its combination of search history, Maps data, YouTube engagement history, and Android device penetration) can credibly build.

NotebookLM and the Knowledge Work Tool

NotebookLM — Google’s AI-powered research and note-taking tool — received substantial updates at I/O 2026 that move it from a consumer productivity tool toward enterprise knowledge management. The enterprise tier, introduced in GA at I/O, allows organisations to deploy NotebookLM on top of internal document repositories, enabling employees to query institutional knowledge the same way they would query a curated research corpus.

NotebookLM’s audio overview feature — which generates a conversational podcast-style summary of a document or research topic — has been particularly successful with enterprise learners who absorb information better through audio than text. The feature is technically trivial (text-to-speech over a structured summary) but commercially clever: it creates a usage pattern that is highly sticky and differentiates NotebookLM from generic AI summarisation tools.

The enterprise NotebookLM play is a direct challenge to Microsoft’s Copilot positioning in knowledge management. Both products do similar things — surface relevant organisational knowledge in response to natural language queries. Google’s advantage is the quality of its foundation model for information synthesis; Microsoft’s advantage is integration depth within the Microsoft 365 data graph. The competition will be decided in enterprise IT evaluation cycles over the next 12-18 months, with data sovereignty configuration and existing vendor relationships the primary decision criteria.

What I/O 2026 Reveals About Google’s Strategic Position

Google enters mid-2026 in a stronger AI position than the conventional narrative — which spent 2023-2024 focused on OpenAI’s lead and Google’s alleged fumbling — suggested. Gemini 2.0 Ultra’s benchmark leadership, Android 16’s deep integration, and the measured expansion of AI Overviews reflect a company that has caught up technically and is executing a coherent commercial strategy.

The existential risk that preoccupied Google’s leadership from early 2023 — that AI search alternatives would erode the advertising revenue base before Google could adapt — has not materialised at scale. Perplexity, you.com, and other AI search alternatives have not taken measurable market share from Google Search. The 40% AI Overviews penetration is Google’s own cannibalisation of its click-through revenue, but it is happening on Google’s terms, at Google’s pace, with Google’s advertising infrastructure capturing most of the value.

The medium-term risk is not displacement but margin compression. A world where AI Overviews handle 70-80% of queries with embedded, lower-CPM ads is a structurally less profitable search business than the pre-AI baseline. Google’s response — growing query volume through better user experience and expanding beyond search into Assistant, Cloud, Workspace, and device AI — is the right playbook. Whether the revenue diversification happens fast enough to offset the core search margin compression is the question that Google’s financial results over the next three years will answer.

I/O 2026 showed a company that knows what game it is playing. Whether it wins that game is a different question.