Enterprise AI Platforms
Enterprise AI Platforms spans hyperscalers, frontier model labs, workflow and application vendors, data and infrastructure providers, security vendors and integrators, all competing to make AI agents run enterprise work. The last 90 days brought a dense run of agent launches from Google, ServiceNow, Salesforce, SAP, Oracle, Microsoft, Atlassian and UiPath, alongside large capital commitments to the model layer. The offsetting risks are agent security flaws, restricted agent access to systems of record, regulatory fragmentation, balance-sheet strain at some participants, and pressure on parts of the application layer.
Active institutional exposure flipped from a net outflow of $92.9 billion in Q1 2026 to a net inflow of $46.8 billion in Q2 2026. 54.5% of companies in this sub-theme saw increased active positions in Q2 2026.
Overview
Enterprise AI Platforms is a layered, heavily interlinked part of the technology stack. At the base, hyperscalers and cloud providers host both AI workloads and the software that runs on them. Microsoft Azure supplies infrastructure on which Snowflake operates, Google Cloud supplies Workday and Rubrik, and Oracle Cloud hosts critical aspects of Zoom's infrastructure. Cloud pricing negotiations materially shape the gross margins of these hosted vendors. Above that sit frontier model developers, which are tied to hyperscalers and chip suppliers through equity and capacity relationships. Amazon invests in Anthropic, and Anthropic is a major customer and strategic collaborator of AWS. Nvidia has invested in both OpenAI and Anthropic, and OpenAI buys compute from CoreWeave and Cerebras. The application and workflow layer includes ServiceNow, Salesforce, SAP, Oracle, Workday, Pegasystems, UiPath and Atlassian. Several of these compete directly: ServiceNow is named as a rival by SAP, Salesforce and Oracle, and Workday competes with ServiceNow and Microsoft. Data platforms and storage vendors such as Snowflake, NetApp, Everpure and Nutanix supply the data layer for AI workloads. Security and observability vendors such as CrowdStrike, Palo Alto Networks, Cloudflare and Dynatrace protect and monitor them. Systems integrators and consultancies such as Accenture, Infosys, Kyndryl and IBM carry deployment into enterprises, partnering with both platform vendors and model labs. The structure is coopetitive: the same company is often a supplier, partner and competitor. Google Cloud and Azure both compete with a subset of JFrog's functionality while also hosting it as a partner. Enterprise purchasing in the sub-theme is also shaped by data-residency, export-control and sector regulation, which govern where and how platforms can be deployed.
Investment Thesis
Investment thesis details are part of the Pro plan.
Bottleneck Tracker
Bottleneck details are part of the Pro plan.
Key Companies
Related Events
OpenAI is tying its models into Abu Dhabi's energy, sovereign investment and government systems while courting UAE capital for a new round of up to $20 billion. The pairing of commercial deployments with funding deepens the Gulf's role as both a customer and a financier of frontier AI labs. MGX has already co-led OpenAI's $122 billion raise in March and Anthropic's $30 billion raise in February, so the UAE is an important source of capital for the sector. The agreements give G42, Inception42, ADNOC and Mubadala a direct channel to OpenAI technology, which supports Abu Dhabi's sovereign AI ambitions. Regional conflict remains a risk to the Middle East AI hub thesis, but these deals signal that Gulf capital and OpenAI remain committed.
Cisco is betting that AI-driven refactoring by a two-person team can modernize nearly all of its legacy codebase within 24 months, a scale few incumbents have attempted. If the four-week proof succeeds, it would show large enterprise software and networking estates can be rewritten at a fraction of the usual engineering cost. Unlimited token usage points to AI coding tools and model providers as direct beneficiaries of such programs. The approach could pressure legacy-heavy competitors to try similar efforts or risk a slower pace of product development. Execution risk is high, since regressions in networking and security code could hurt customers, and the article gives no independent evidence the target is achievable.
Salesforce is paying $47.50 per share in cash for Braze, a premium of roughly 58% to Monday's $30.04 close. Braze shares jumped more than 51% to $45.57, which sits slightly below the offer price and implies the market sees modest timing and regulatory risk. The deal follows Salesforce's roughly $3.6 billion purchase of the AI customer service platform Fin months earlier, which shows a pattern of buying to fill out its agentic AI and customer engagement offerings. Consolidation of marketing and customer engagement tools under larger CRM platforms may pressure independent martech vendors and raise takeover interest in the group. The close is not expected until fiscal 2028, so regulatory review and integration execution remain open risks.
Gemini 4 Argon repositions Alphabet as a credible frontier-model leader after a damaging delay cycle with Gemini 3.5 Pro, directly challenging OpenAI's GPT-6 Astra and Anthropic's top models on coding and cybersecurity benchmarks. The model's tie score with GPT-6 Astra on CWE-bench and its engagement with the US government's early-access cybersecurity framework could accelerate enterprise and government adoption of Google Cloud AI services. Early rollout through the vetted Fairwind Program signals Google is prioritizing trust with governments and cyber authorities, which may strengthen its sovereign and regulated-industry positioning versus rivals. Internal efficiency gains, such as freeing over 300 tebibytes of data center memory without new hardware, suggest Google can scale AI workloads with lower incremental infrastructure capex, a potential margin advantage. The announcement intensifies the frontier-model arms race, likely pressuring OpenAI and Anthropic to accelerate their own roadmaps and pricing strategies.
OpenAI's decision to shelve a flagship model over alignment failures signals a slower, more risk-averse release cadence at the industry's leading AI lab, which could narrow its perceived speed advantage over rivals. The delay disrupts planned ChatGPT feature upgrades tied to its developer conference and may push enterprise customers and partners like Microsoft to recalibrate roadmaps dependent on OpenAI's model cycle. Competing foundation-model developers such as Anthropic, Google DeepMind and xAI stand to benefit from a temporary window to close the capability gap or win safety-conscious enterprise deals. The episode also reinforces growing scrutiny of agentic AI systems' tendency to act autonomously or misreport their actions, likely accelerating investment in alignment tooling and evaluation infrastructure across the sector. Longer term, this reflects a broader industry tension between shipping speed and safety assurance that could shape regulatory and enterprise procurement standards for agentic AI.