AWS网络指标数据：2026年第二季度中期警示信号浮现？

AWS Web Metrics Data: Mid-Q2'26 caution signal emerging? Our regular web metric updates over the last few quarters showed exceptionally strongdemand signals — arguably the strongest since COVID. The strength has translated intoreported results, with revenue growth acceleration in AWS as AI workloads slowly ramped,and cloud consumption linked names like Datadog, Snowflake, and Cloudflare. That said,we want to flag a trend that we are watching closely:growth flattened unusually over thelast four weeks (+ start of this week) relative to the pattern seen over the prior threeyears(Exhibit 2). Reminder: this is alt data with good historic fit, but for a variety of reasonsit could stop being predictive. Fundamental explanation remains important. Mark Shmulik+1 917 344 8508mark.shmulik@bernsteinsg.com Peter Weed+1 917 344 8390peter.weed@bernsteinsg.com Luwei Yang+1 917 344 8342luwei.yang@bernsteinsg.com Limited Q2 revenue impact expected.It is important to remember that our dataeffectively operates on a one-quarter lead relative to reported financials. What we aremeasuring is closer to incremental ARR added in the quarter, whereas reported resultsreflect postpaid billing and revenue recognition from existing ARR. As a result, Q2 revenueshould largely reflect the strength we saw in Q1 web metric signal (Exhibit 1)— we thinkQ2 could show another 100 bps acceleration. Deeksha Pandey+1 917 344 8447deeksha.pandey@bernsteinsg.com Wenhuan Chang+1 917 344 8546wenhuan.chang@bernsteinsg.com Q3 non-AI revenue impact likely limited.Q2 started strong, so if the recent impliedweakness reverses in June, Q3 would still have a “normal” Q/Q growth. Note there may be amodest mathematical Y/Y headwind to non-AI growth as Q3’25 was strong. Armin Hadavi, CFA+1 917 344 8463armin.hadavi@bernsteinsg.com Q4 could be an issue.Where the setup becomes more challenging is if the currentflattening persists into Q3 web metric data.If the trend remains subdued, the resultwould be a pronounced Y/Y growth headwind to non-AI consumption in Q4. What’s driving the softness.We asked leaders of top software companies, large growthsoftware PE portfolios, as well as consultants working with enterprise buyers andthecommon thesis we heard: hyperscalers are running into capacity constraints withincremental compute heavily allocated toward frontier AI labs,leaving less availablesupply for other buyers (e.g., enterprises, software vendors). More narrowly, from a verydeeply integrated source within product team buyersa couple added thesis emerged: 1)Mythos induced cyber urgencycreates a HUGE backlog of issues to address. Companiesare sprinting to do this, as opposed to working on next incremental new workload.2)tokenmaxxing hangoveras some product teams attempt to moderate spending growthon coding agents and AI copilots. After a period of rapid adoption, there is some effort toreset and impose discipline. Regardless of the driver, the shift is notable and raises a warning signal that wewill keep watching.A softer consumption backdrop in Q3+ could impact demand fromtraditional enterprise buyers and cloud software vendors, while less problematic for the“Born-in-AI” cohort (i.e., AI-linked trajectory for AWS and Datadog). Could AWS benefit?In practice, supply constraints point to a more competitiveenvironment for securing capacity to bring new workloads into production, which could giveAWS pricing leverage. Similarly, if available capacity has shifted towards higher margin AIworkloads, then AWS may also find its financials in a better position. BERNSTEIN TICKER TABLE INVESTMENT IMPLICATIONS No change to models, price targets, or recommendations. Amazon: we maintain our Outperform rating and $315 PT. Datadog: we maintain our Outperform rating and $180 PT. Twilio: we maintain our Market-Perform rating and $183 PT. Cloudflare: we maintain our Market-Perform rating and $136 PT. DETAILS EXHIBIT 2:The strong acceleration continued in early Q2with the web metric data performing better than theprior 3 years until the first week of May. But the trendflatlined in May, underperforming the trends we saw in2023 and 2024. EXHIBIT 1:2026 had a strong start with the AWS webmetric data accelerating in January and maintaining thestrong level throughout Q1. WHY DO WE CARE ABOUT AWS WEB METRICS DATA SSO DATA SERVES AS A LEADING INDICATOR FOR AWS EX-AI REVENUE As we have shared over the past few years, we have observed a strong correlation between -1 quarter (last quarter)engagement with AWS' single-sign-on ("SSO") page and current quarter revenue growth for AWS. This makes sense as theamount of human interaction with the SSO can't really oscillate that much due to personal work preference, and instead itmore likely represents the portion of people actually tasked with doing projects on AWS. And because the amount of workany individual person can do (and is expected to deliver) remains pretty consistent over time, and their work turns into similaramount of rev