Software Stocks Slide as Snowflake and Datadog Fall While Broader Tech Holds Steady

The market split on Wednesday. Big-tech indexes barely budged. High-multiple software did not get the same courtesy.Snowflake fell about 4% to around $306 ahead of its fiscal second-quarter 2027 report. Datadog dropped about 6% to around $211 with no earnings on the calendar and no fresh company headline. Cloudflare also slid about 4%. CrowdStrike, Fortinet, ServiceNow, and other software and security names were weaker too.The Nasdaq 100 finished slightly higher. The S&P 500 was roughly flat. That gap is the story. Money left a corner of technology, not technology as a whole.Snowflake then reported after the close and beat Wall Street. Revenue came in at $1.55 billion, up about 35% year over year, against estimates near $1.48 billion. Adjusted earnings were $0.62 a share versus a $0.45 consensus. Product revenue rose 37% to $1.49 billion. Management raised full-year product-revenue guidance. Shares firmed in extended trading.The daytime selloff still matters. It showed what investors feared before they heard the numbers — and what they are still doing to the rest of the software group.Rotation, Not One Broken CompanyIf only Snowflake had fallen, the easy explanation would be earnings nerves. A stock up roughly 46% year to date invites profit-taking into a print. Options implied a large move either way. That is normal.Datadog ruins that tidy story. It was not reporting. It still fell harder than Snowflake. Cloudflare was not reporting either. When three of the year’s winners decline together while the Nasdaq holds up, traders are not reacting to a single 8-K. They are reducing exposure to a style: expensive, consumption-based, AI-adjacent software.That style had a huge first half. Datadog was up about 65% year to date before the latest slide. Cloudflare was up about 45%. Those cushions make selling easier. They also make the group sensitive to any hint that growth is slowing or that customers are concentrating spend on a few giant AI platforms instead of a long list of SaaS tools.The Hangover From AugustDatadog already taught the market a painful lesson on August 6. The company beat on revenue and earnings and raised full-year guidance. The stock still crashed. Management said a very large AI customer would use less starting in the third quarter and built that cut into the outlook.That single disclosure reframed the sector. Consumption businesses make money when customers burn more compute, more logs, more queries. They lose multiple points when one whale throttles usage. Investors began asking a rude question: is AI a demand engine for software, or a customer-concentration risk?HubSpot and other names added to the unease around pricing and AI attach. MongoDB later beat and raised and still sold off as traders focused on infrastructure costs. Palo Alto Networks beat and guided higher and still dropped double digits after a long run. The pattern is familiar. Good quarters are not enough if the multiple assumed perfection.By early September, Datadog was down about 22% in a month even as some peers held up better. Wednesday’s 6% drop sat on top of that reset.Why Software Can Fall When Nvidia Does NotThe session’s other headline was strength in parts of the AI-hardware complex. Nvidia and deal headlines helped the broader tape. Software did not tag along.That divergence is becoming a feature of 2026. Chips and data-center suppliers get paid when companies build AI. Application software gets paid only if those companies keep buying seats, usage, and extra modules. If a customer spends the budget on GPUs and a frontier model, the observability bill, the data-cloud bill, and the security bill have to fight for what is left.There is a second fear: AI tools may let firms do more with fewer software vendors. That thesis has been wrong before and may be wrong again. It is still moving prices. High free-cash-flow multiples leave no room for a long debate.Geopolitics and rates did not help risk appetite at the start of the week. Middle East tensions and higher bond-yield talk made long-duration growth stocks an easy source of cash. Software is long duration. Hardware with near-term AI orders has been treated as a different animal.What Wednesday’s Tape Was PricingInto the close, Snowflake’s decline was a hedge against disappointment. After the close, the report argued the opposite: product growth accelerating, AI consumption contributing, remaining performance obligations still expanding, margins wider than a year ago.That does not automatically end the sector selloff. One beat can lift a name overnight and leave Datadog, Cloudflare, and the software ETF heavy the next morning. Markets have been willing to cheer a print and still fade the group.The clean read from Wednesday is therefore two-layered.Layer one: traders used a quiet Nasdaq day to cut winners in software.Layer two: Snowflake’s fundamentals, at least for one quarter, did not confirm the panic. They confirmed demand for a data platform that sits next to AI workloads.Both can be true. Positioning can be ugly while the business is fine. The business can be fine and the multiple can still compress.Consumption Models Live and Die on GuidanceSnowflake, Datadog, and MongoDB do not collect the same kind of rent Salesforce collects. Seat-based SaaS is lumpy but visible. Usage-based cloud is honest and violent. A customer can double consumption in a quarter or slash it when a training cluster goes quiet.That is why Datadog’s large-customer comment still hangs over the group, and why Snowflake’s product-revenue acceleration was the number that mattered after hours. If AI workloads keep expanding inside the data cloud, the consumption model is the bull case. If a handful of AI labs and hyperscalers dictate the slope of usage, the model is a lottery ticket.Investors will parse Snowflake’s commentary for the mix of new logos versus existing-customer expansion, and for whether AI products are incremental or just a new label on old queries. They will watch Datadog for any update on that large customer. They will watch Cloudflare for whether internet and security traffic growth can outrun a multiple reset.Not a Broad Tech CrashIt is important not to over-write Wednesday into a 2022 sequel. The S&P 500 and Nasdaq 100 were not breaking down. The damage was concentrated in software and some cybersecurity names that had already run. That is rotation.Rotation can last. If funds decide the AI trade belongs in semiconductors, power, and a few platforms, software can stay a source of funds for months even when earnings are decent. It can also snap back the first time a cluster of prints looks like Snowflake’s: faster growth, higher guide, proof that enterprises are still writing bigger checks.The honest stance is to separate the stock from the sector. Snowflake’s quarter answered the company question. It did not fully answer the sector question. Datadog’s silent 6% drop is still sitting there as evidence that the question remains open.What Comes NextNear term, watch whether Snowflake’s after-hours bid holds in regular trading and whether Datadog and Cloudflare bounce or keep leaking. Watch the iShares software ETF for whether this is a one-day trim or another leg down in a year-long argument about AI and SaaS.