Anthropic - My Portfolio Killer

After Anthropic revealed what Mythos could do, the stock market reacted immediately.

And not in a subtle way.

Cybersecurity stocks sold off hard.

CrowdStrike fell about 6.6%.
Zscaler dropped roughly 11%.
SailPoint was down 8.9%.
Palo Alto Networks fell 4.7%.
Fortinet dropped 3.8%.

Investors were not treating Mythos like just another AI product launch.
They were treating it like a warning shot at parts of the existing cybersecurity business model.

That reaction tells you something important.

The market was not saying, “Cybersecurity is over.”

It was saying something more specific:

If AI can now discover serious vulnerabilities much faster than human teams can process them, then which cybersecurity firms are actually protecting customers… and which ones are mainly monetising the slowness of the old system?

To understand what Mythos just did to cybersecurity firms, you first have to understand how a lot of cybersecurity firms actually make money.

They do not make money by selling “safety” in some abstract way.

They make money from a few very specific value pools:

1. Prevention
Protect the endpoint, the network, the identity layer, the cloud workload.
Usually sold as subscriptions. Per user, per endpoint, per workload, per module.

2. Detection and visibility
Collect logs. Monitor behavior. Raise alerts. Correlate signals.
This is the world of dashboards, SOC tooling, threat intel, SIEM, MDR.

3. Discovery and prioritization
Find vulnerabilities. Scan code. Flag misconfigurations. Rank what matters.
This is where a lot of vulnerability management, appsec, posture management, and assessment tools sit.

4. Response and recovery
When something breaks, humans come in.
Incident response. Forensics. Remediation. Advisory.
Often billed as services, retainers, or time-and-materials.

Now here is the important part.

For a very large part of the industry, the real money was never just in “finding problems.”

It was in managing the queue.

Finding the issue.
Labeling it.
Prioritising it.
Escalating it.
Routing it internally.
Creating workflow around it.
Selling comfort around it.

That was valuable because the digital world became too complex for human teams to keep up with.

So companies paid for help.

Help seeing the problem.
Help understanding the problem.
Help deciding which problem mattered first.

That has been a huge value pool.

And this is exactly where Mythos creates discomfort.

Because if Anthropic is right, and Mythos can find serious vulnerabilities across major systems at a level that changes the speed and depth of discovery, then one big part of cyber economics starts looking weaker. Anthropic says Glasswing exists to secure critical software for the AI era and is giving defenders early access to Mythos Preview because of its capabilities.

Not cybersecurity itself.

The old premium on human-speed discovery and triage.

That is what the market reacted to.

The fear was not:

“Oh no, companies won’t need cybersecurity anymore.”

The fear was:

“What happens to cybersecurity firms whose core value is helping human teams process weakness after weakness, when AI starts finding those weaknesses much faster than the humans can even read the tickets?”

That is a very different question.

Think about it like this.

Imagine you own a hospital.

For years, your business model is built around the emergency desk.

Patients come in.
Someone checks symptoms.
Someone categorises urgency.
Someone decides where the case goes.
Someone moves the file forward.

Now suddenly a machine shows up that can identify serious conditions across the whole building far faster than your front desk can process them.

The bottleneck is no longer diagnosis.

The bottleneck becomes:

Can you treat?
Can you isolate?
Can you operate?
Can you fix the root cause before the next patient comes in?

That is what Mythos may do to parts of cybersecurity.

It shifts the bottleneck.

From finding
to fixing

From visibility
to remediation

From alerts
to action

From human workflow
to machine-speed defense

And whenever the bottleneck shifts, value pools shift with it.

So where does value likely get weaker?

In my view, the most vulnerable pools are the ones built mainly around:

• surfacing large volumes of issues

• ranking those issues

• routing those issues

• helping analysts manually work through those issues

• selling another dashboard to explain why the backlog still exists

Because if AI-native systems can find deeper issues faster and more cheaply, then “we help you manage the pile” becomes a less powerful pitch.

That is where pricing pressure begins.

That is where margins get questioned.

That is where investors start asking whether some tools are actual protection…

…or just expensive workflow software for a slower era.

But the story does not end there.

Because while one value pool weakens, another gets stronger.

The next big pools in cybersecurity may be:

1. Automated remediation
Not just spotting the flaw, but helping fix it.

2. Patch validation and safe deployment
Making sure the fix works without breaking production.

3. Containment at machine speed
Identity controls, segmentation, runtime protection, rapid isolation.

4. AI-native SOC and response
Not more alerts. Faster action.

5. Code hardening and secure software pipelines
Especially for critical infrastructure and open-source dependencies.

6. Managed AI defense
Because most companies still will not have the internal talent to run all this themselves.

So no, I do not think Mythos is a cybersecurity market killer.

If anything, cyber spend may go up.

The attack surface is getting larger.
The pace of exploitation is getting faster.
And the average enterprise is nowhere near ready.

But I do think Mythos could be a legacy cyber workflow killer.

Or said more bluntly:

A killer of the part of the market that made money from the fact that humans were too slow.

That is the real shift.

The winners in cybersecurity now may not be the firms that are best at telling you what is wrong.

They may be the firms that can help you fix what is wrong before AI-assisted attackers move first.

That is where the next value pool will sit.

And that is why the market got nervous.