CrowdStrike ships a model trained on AI attacks as SentinelOne launches a rival
Fal.Con Europe opened in Barcelona with the first SafeMind-trained model due in Falcon this week, days after CrowdStrike reported an agent-driven campaign that hunted for machines without its sensor.
Richard Tang · 3 min read
Fal.Con Europe opened in Barcelona on November 2, sold out and running to November 4 1, and it is the week CrowdStrike's−0.37% — CrowdStrike, down 0.37 percent today boldest research turns into product. The first SafeMind-trained defensive model, built in the company's attack-emulation factory, ships into the Falcon platform this week 2. With the stock up 114% this year 3, the question is narrow: does in-house AI defense convert into sellable product before the AI labs intercept the same budget?
The threat it trains against routes around the sensor
CrowdStrike's October 29 report is the demand proof 4. It attributes to VAPOR PANDA, a crew it links to China, a campaign that ran the open-source ARTEX agent framework against more than 200 organizations 4. It counts confirmed intrusions at nine of them 4. The recovered transcript carries the detail that matters for SafeMind: a human operator told the agents to find Windows machines in the domain without Falcon 4. Attackers now hunt for the machines the sensor is not on, which is a threat to the endpoint model itself.
SafeMind answers by construction. Red Tempest attack models, built on Nvidia's+1.00% — Nvidia, up 1.00 percent today Nemotron plus safety-stripped open-weight models, run loose across 40,000 to 60,000 emulated corporate networks, and their attack traces train Blue Solano defenders that harden those environments on their own 2.
What actually sells this week
Day one brought a CoreWeave partnership putting Falcon protection and managed security on cloud AI workloads 5. On November 2 CrowdStrike also expanded Project QuiltWorks across Europe, adding Atos, BT, Deutsche Telekom Security and Telefónica among fifteen partners who turn AI-found vulnerabilities into fixed ones 6.
QuiltWorks packages frontier models from OpenAI$1.18T — OpenAI, private, latest valuation $1.18T and Anthropic$350B — Anthropic, private, latest valuation $350B with CrowdStrike intelligence, which puts the labs' models inside CrowdStrike's delivery machine rather than selling around it 6. Later phases open QuiltWorks as trusted access to the Red Tempest and Blue Solano models themselves, and sell the emulation environment, including air-gapped installs for governments 2.
Three camps want the same budget
SentinelOne−4.47% — SentinelOne, down 4.47 percent today launched Singularity Agentic Detection and Response this week, its product chief calling traditional endpoint tools flying blind against agents 7. Gartner+3.52% — Gartner, up 3.52 percent today puts spending on AI-security platforms at $3 billion this year 3. By 2028 it expects $7.7 billion 3. OpenAI's finance chief calls security a huge business, and Anthropic's IPO filing names it one of the company's most important near-term opportunities 8. CrowdStrike's counter is distribution, and it is already paying: Falcon Flex−0.28% — Flex, down 0.28 percent today annual recurring revenue has passed $2.2 billion 3.
The trade
Deutsche Bank upgraded CrowdStrike to Buy with a $300 target on November 2, the day the conference opened 9. The shares had closed 4.1% lower on Friday 3. Analyst targets now span $265 to $320 9.
CrowdStrike analyst targets now run from $265 to $320 after Deutsche Bank's upgrade
Data
| Value | |
|---|---|
| Freedom Broker | $320 |
| BTIG | $312 |
| Deutsche Bank | $300 |
| Citigroup | $295 |
| Mizuho | $275 |
| KeyBanc | $265 |
The capability is real, the distribution is enviable, and the labs already sit inside CrowdStrike's machine. What turns the story into revenue is whether the models shipping this week measurably harden customer fleets, and that test shows up in renewal data, not keynote applause. Meeting AI-powered attacks with better AI defense is the right answer, which makes conversion speed, not the technology, the thing to judge from here.
Get Security by email, free
The breaches, flaws and security vendors AI is putting to the test. In our daily email, every weekday.
