The main line of AI investment has entered its second phase! "Turning Tokens into Cash Flow" dominates the allocation logic. Dynatrace (DT.US) is riding the super wave.

Palantir (PLTR.US), known for its "myth of AI applications," has shown strong performance since July, indicating that global investors are gradually shifting their AI valuation anchor from "how many GPUs one has and the list of beneficiaries from the AI infrastructure boom" to "whether tokens can be converted into actual revenue, profit, and verifiable productivity." Recently, Morgan Stanley released a research report that raised the stock rating of Dynatrace (DT.US), which supports the application monetization phase termed the "digital systems control tower," from "hold" to "overweight," increasing the target price from $58 to $65. The core logic behind this change is that the company can generate sticky subscription revenue amid enterprise AI cloud adoption, the increasing complexity of AI inference workloads, and vendor consolidation. Recently, a notable relative rotation has occurred within U.S. tech stocks, shifting from "single-line AI theme trading in AI computing power infrastructure" to "spreading towards AI application software monetization," but it cannot yet be said that funds have completely withdrawn from the AI computing theme and semiconductors to focus solely on AI application software; this ongoing transition from AI computing infrastructure investment themes to software stocks does not imply that software stocks are rising indiscriminately. After its earnings report on August 4, Palantir's stock surged by nearly 29.5% in a single day. However, as of August 24, the iShares Software ETF (U.S. ETF Code: IGV) is still down 3.01% year-to-date. A more accurate trend indicates that a structural re-evaluation centered on profit realization is occurring within software stocks, rather than the entire software sector entering a bull market. As global funds pivot from the first phase of AI-related investments focused on GPU/HBM/AI data center core hardware bottlenecks to the second phase of application layer winners that can translate tokens into increasingly robust enterprise productivity, revenue, and cash flow, future valuation differentiation is likely to intensify: software companies with exclusive data, workflow entry points, closed-loop agent intelligent workflow execution, and clear ROI may be revalued, while traditional SaaS companies that risk becoming commoditized through basic models may continue to face pressure. The torrent of token inference requires a "digital control tower"! Morgan Stanley maintains a bullish outlook on Dynatrace. Dynatrace's stock has surged approximately 50% since hitting its year-to-date low point in April. Dynatrace is not a directly consumer-facing AI application but serves as a "digital systems control tower" supporting the monetization phase of AI application software: as enterprises integrate large models, inference Co-pilots, and AI agents into production processes, the complexity of application call chains, token consumption, latency, error rates, and infrastructure dependencies escalates dramatically. Observability has been upgraded from traditional operation and maintenance tools to a key software layer that ensures the reliability, cost efficiency, and business continuity of the entire AI system. Dynatrace is a software company that encompasses enterprise-level unified observability and software application security. Its core product is not a standard "system monitoring tool," but rather a real-time diagnostic and automated decision-making platform for complex digital systems: by uniformly collecting metrics, logs, distributed traces, user behavior, and business data, it helps enterprises identify performance bottlenecks, predict failures, analyze root causes, and automatically execute fixes. The company primarily charges through Software as a Service (SaaS) subscriptions, regular licenses, and operational support, with the Dynatrace platform being its main source of revenue; as of March 31, 2026, the company had approximately 4,100 customers spread across over 110 countries. Its business scope encompasses six areas: application performance and cloud-native observability, log and infrastructure monitoring, digital experience monitoring, application security, business observability, and AI observability focused on large models, generative AI, and intelligent agents; the latter can monitor token costs, inference delays, model quality, and abnormal calls across AI GPU computing clusters, base models, vector databases, semantic caching, and agent orchestration frameworks. Dynatrace's more accurate investment positioning is as the "digital systems control tower" of the AI application erait does not directly produce large models or tokens, but rather derives high-margin subscription revenue from enterprise cloudification, the complexity of AI workloads, and vendor integration; in the first quarter of the 2027 fiscal year, its annual recurring revenue (ARR) reached $2.136 billion, with a year-on-year growth of 17%, and subscription revenue alone amounted to as much as $530 million. Therefore, Dynatrace represents the transition of funds from purely focusing on AI computing industry hardware bottlenecks to "AI application foundational software" logicit may not directly generate token revenue but is responsible for ensuring that tokens can be stably, economically, and auditably transformed into business outcomes. Morgan Stanley's bullish view on Dynatrace is based on a quadruple resonance of "industry recovery + expand renewal cycles + platform expansion + margin improvement": the demand for observability is at its healthiest state since 2022, the subscription renewal foundation for the Dynatrace platform is rapidly expanding, and log and AI agent-driven inference workloads along with autonomous operation and maintenance products are significantly increasing client spending share. The firm expects growth to continue exceeding 20% and has raised its rating from "hold" to "overweight," with the target price raised from $58 to $65. Following the upgrade of Dynatrace (DT.US) from "hold" to "overweight," along with a simultaneous target price raise by Morgan Stanley, the company's stock price rose by over 2% by the close of U.S. markets on Tuesday, ending at $50.07, indicating substantial potential for further appreciation. Morgan Stanley has also raised Dynatrace's target price from $58 to $65. From Morgan Stanley's perspective, Dynatrace's observability platform demonstrates critical value in the AI inference erait is a software tool that analyzes data to track the performance and operational patterns of complex digital systems and assists in resolving software issues. Led by analyst Sanjit Singh, Morgan Stanleys research team stated: "In one of the healthiest observability market environments in years, Dynatrace is poised for accelerated growth driven by a rapidly expanding renewal base and an increasingly wide product suite capable of winning platform integration deals. The acceleration of growth to over 20%, enhanced market positioning for AI applications, and margin expansion should propel its free cash flow valuation multiple, thereby supporting our $65 target price." Analysts unanimously pointed out that driven by strong observability demand, renewal of Dynatrace Platform subscriptions (DPS), and platform expansion, they believe the company has a pathway to achieve sustained, long-term growth of over 20% while continuously expanding margins. Singh and his Morgan Stanley analyst team indicated that with the acceleration of cloud computing growth, digital transformation, and future large-scale adoption of AI agents by enterprises, the demand for enterprise-level observability is currently at its healthiest level since 2022. The analysts highlighted that the company's cross-platform expansion into logs, AI, and autonomous operations has increased its chances of capturing a greater share of client IT spending and strengthened its market position as an integrated platform vendor. The focus is gradually shifting from "who leads the deployment and benefits from building the largest GPU data centers" to "who can convert tokens into sustainable cash flow." The valuation anchor for the AI investment wave that began at the end of 2022 is gradually upgrading from "capex scale" to "capital return efficiency," and this process is acceleratingi.e., the first phase of the AI investment frenzy centered entirely on "who leads the deployment and benefits from building the biggest GPU data centers," while the current second phase focuses on "who can convert tokens into sustainable cash flow." The super bull market surrounding AI is shifting from "buying chip stocks" to "buying AI workflows," meaning that the current market is re-pricing the main line of AI bull market investments from "who benefits from ever-increasing AI capital expenditures" to "who can most quickly convert computing power into ARR, margins, and free cash flow." This latest rotation favors software companies with embedded critical business processes, high renewal rates, data barriers, and the ability to monetize intelligent agents. The second phase of AI investment questions whether these computing powers can achieve higher utilization rates, lower unit token costs, and sustainable revenues and free cash flow. The decline in inference prices enables enterprises to deploy more task cycles, tool calls, and multi-agent workflows, while demand growth may exceed the reduction in unit prices, creating a total computing power expansion analogous to the Jevons Paradox. Wall Street financial giant Goldman Sachs Group, Inc. predicts that AI agents could drive an astounding 24-fold increase in token consumption by 2030, suggesting that the demand for artificial intelligence computing power has not cooled at any degree, but surplus marginal gains are spreading to cloud platforms, AI applications, and key software infrastructures capable of embedding tokens into enterprise workflows, converting them into productivity, and establishing a charging loop. Triple pressure from credit, supply chains, and policies is forcing stock markets to revisit the AI capital expenditure wave with a bond investor's discipline. Morgan Stanley forecasts that the size of global AI-related bond issuance will approach $570 billion by 2026, with $236 billion reached by the end of May. The four major cloud vendors' spending is estimated to be around $700 billion for the year; when credit default swap (CDS) spreads, collateralized obligations, and off-balance-sheet financing costs rise, "forward income stories" must undergo solvency and cash flow coverage tests. The AI computing theme trading has not come to an end; rather, it is bidding farewell to indiscriminate valuation expansion: platform cloud computing giants and AI application software that are flush with cash and benefit from surging token consumption, alongside suppliers of storage chips and the core segments of data center infrastructure that show significant divergence in stock prices and EPS trajectories, still have opportunities. Dynatrace precisely embodies the intermediate-layer opportunity in this round of "token cash flow realization": it does not directly sell AI agent terminal applications, but rather monitors model latency, token costs, call quality, and root causes of failures through AI observability, application performance management, log analysis, security, and autonomous operations, transforming enterprise AI from experimental projects into reliable, controllable, and scalable revenue-generating production systems. Morgan Stanley predicts its annual recurring revenue (ARR) is expected to continue growing by over 20%, with the platform's renewal customer base expanding by about 50% compared to the previous cycle, highlighting that, in Morgan Stanley's view, the market is willing to pay a premium for software platforms where "enterprise AI agent penetration rates can renewals, margins, and free cash flow."