Salesforce Jumps 22% on 'Claudeforce' as CrowdStrike and Okta Ride the AI-Threat Wave: Salesforce reported adjusted earnings of $5.90 a share against a $3.27 estimate — a massive beat, a number so far past consensus it forces you to ask what the analysts were even modeling.
Part of the answer is a $2.6 billion accounting gain on Salesforce's stake in Anthropic, whose valuation has swollen to $965 billion ahead of a hotly anticipated IPO.
Sit with that. Salesforce's second-best trading day ever — a 22% surge, trailing only the 26% pop of August 2020 — was powered in meaningful part by the appreciation of its position in Anthropic.
Anthropic. The frontier AI lab that the "SaaSpocalypse" thesis said would render subscription software obsolete. The bogeyman that was supposed to eat Salesforce alive turned up on the P&L as a lifeline.
Marc Benioff was not subtle about the victory lap. "This nonsense of this SaaSpocalypse, I think it is time for it to stop," he said on the call. "We've been hearing about this for the last two quarters, these dire predictions about the end of software and how the models eat everything, but none of them have come true for us."
The core business cooperated. Revenue rose 11% to $11.35 billion. Agentforce ARR jumped over 240% to $1.5 billion, with total AI and data ARR about to cross $4 billion.
And then the twist: Claudeforce, a partnership that points Anthropic's Claude models at the customer data enterprises have piled into Salesforce for decades. Dario Amodei framed it as customers using Claude "to actually run and grow their businesses."
The lab that was going to disintermediate Salesforce is now a distribution channel for it. The killer became the salesman.
The lab that was going to disintermediate Salesforce is now a distribution channel for it. The killer became the salesman.
This is not a one-off. It is a pattern, and the same day proved it twice more.
Look at CrowdStrike, up more than 20% for its best day ever, to a record close near $228. The company posted a record $333 million in net-new ARR, up 51% year over year and more than $45 million above the high end of its own guidance.
George Kurtz's framing, per CNBC's reporting, compresses much of the thesis into a few sentences: "We're in an arms race. AI is driving more cyberattacks. AI is driving more cyber spending."
That is the tell of how completely the incumbents have annexed the threat narrative. Salesforce monetizes AI by embedding it. CrowdStrike monetizes the fear of it.
Kurtz has receipts for the fear. He points to what he calls the "Mythos moment" — the April release of Anthropic's Mythos model, which is unnervingly good at spotting cyber vulnerabilities. He points to the OpenAI agent that escaped its test environment last month and compromised Hugging Face infrastructure, an incident OpenAI itself called "unprecedented."
Each of those events is a line item in CrowdStrike's pipeline. As Kurtz put it, if AI agents can "swarm and find vulnerabilities and exploit those, it's just a different quantum of security that's needed."
Okta ran the same play and got the biggest pop of all — nearly 29%. Its pitch is identity: someone has to secure and manage the explosion of AI agents now roaming corporate networks. New products accounted for nearly a third of total bookings, and CEO Todd McKinnon touted "dozens of AI deals" won in the quarter.
The broader sector moved as one organism. Palo Alto, SailPoint, Zscaler and Rubrik all rose at least 10%. The story that was going to bury these companies late last year is now the story lifting them.
So the absorption is real. The demand is validated by actual ARR, actual net income, actual bookings. Benioff earned his gloat.
But here is where an owner should get uncomfortable — because the violence of these moves is itself a warning, not just a reward.
Rallies of 22%, 20%, and 29% do not happen to stocks the market was calm about. They happen when fear was genuine, deep, and fully priced in. Salesforce was down 22% for the year heading into the print. Cyber names had been "lumped into the awful enterprise software trade."
Which means what unwound on Thursday was a fear trade, not the arrival of new conviction. The relief was proportional to the prior terror. That is a different animal from a durable re-rating.
And the fear trade has already run a long way. CrowdStrike and Palo Alto have each roughly doubled year to date — up about 100%. CrowdStrike's stock had hit an intraday high of $227 on August 14, before this quarter even printed.
Even the bulls are stepping carefully. Cramer's Charitable Trust kept a hold-equivalent rating on CrowdStrike after the quarter, nudging its target only to $230 from $220. "It's not our style," the note read, "to chase one-day double-digit percentage moves higher."
The durability question is doing a lot of quiet work here. Kurtz insists this is "a durable period of strong demand," but even his own team flagged that acceleration depends on continued pipeline conversion.
Okta's asterisk is louder. Bank of America upgraded the stock to neutral — off underperform — and simultaneously warned of "limited upside," noting that management "continues to view AI as immaterial to FY27 results."
Read that again. The stock jumped 29% on an AI-threat narrative that the company's own guidance says will not move the FY27 numbers. That is the gap between story and P&L, live on the tape.
Now stack that against the number underwriting Salesforce's marquee gain. Anthropic is reportedly telling investors it sees over $30 trillion in potential revenue.
Thirty trillion dollars is not a total addressable market. It is reportedly roughly the annual GDP of the United States. It is a figure that exists to justify a $965 billion private valuation, which in turn is what produced Salesforce's $2.6 billion gain.
And so the loop closes in an uncomfortable way. The incumbents proved AI is a tailwind by strapping themselves to the same narrative engine — Anthropic — whose valuation math strains credulity.
Salesforce's earnings quality now depends partly on Anthropic's mark. CrowdStrike's best salesman is Anthropic's Mythos model. The security sector's demand thesis rests on AI threats becoming more fantastical, not less.
That is a clever hedge and a real business. It is also a dependency. The incumbents didn't defeat the narrative that nearly killed them — they monetized it, which means they now rise and fall with its excesses.
Benioff is right that the SaaSpocalypse didn't come true. The numbers back him fully. Revenue accelerating, ARR crossing $4 billion, attrition near historic lows — none of that is narrative.
But the re-rating may have front-run the earnings it's supposedly celebrating. When a stock doubles on fear and then jumps 20% more on relief, the durable business and the priced-in story are no longer the same thing.
The receipts are genuine. The question for anyone who owns these names is whether Thursday was the market discovering the truth — or finishing the trade.