AI’s Next Bottleneck Isn’t Chips. It’s Cybersecurity.

Dear readers,

Yesterday we wrote about a market forced to separate cash machines from cash burners now that money costs five percent again. But there’s a second rotation underway that has nothing to do with the Fed — and everything to do with the fact that the smarter AI gets, the more exposed it makes everyone around it. Cybersecurity is turning from a compliance line item into a survival cost for the machine economy, and the spending is happening regardless of what Jerome Powell does next.

A “Tsunami of Patching”

Citigroup CEO Jane Fraser put it starkly this week: the financial industry is bracing for a global “tsunami of patching,” triggered by frontier AI systems like Anthropic’s “Mythos.” The threat isn’t theoretical — the World Economic Forum now finds that 87% of companies rank AI-related vulnerabilities as their fastest-growing risk. Wall Street is split on what to do about it. Goldman Sachs CEO David Solomon has urged the industry to slow down; Nvidia’s Jensen Huang has dismissed that kind of caution as “doomsday narratives” and kept pushing the accelerator. Anthropic and Accenture just settled the argument in practice, announcing a joint $2 billion investment in embedded AI security analytics. The takeaway for investors is blunt: the AI rollout isn’t slowing down, so the security layer underneath it has to scale at the same pace — and that’s a multi-year tailwind, not a quarter’s trade.

Wall Street Is Already Repricing the Winners

That dynamic is showing up in analyst math, not just in headlines. Morgan Stanley just raised its price target on Palo Alto Networks from $394 to $410, keeping it at Overweight. The firm estimates that roughly 1.5% of all enterprise AI capital spending is now flowing directly into AI security — and, contrary to a market consensus that worries about share losses, Morgan Stanley actually expects Palo Alto to gain around 75 basis points of market share over the next two years. The stock, up 99% so far in 2026, has recently traded around $317. For investors hunting for the picks-and-shovels play on the AI buildout, the cybersecurity leaders look like the ones actually cashing the check.

When Machines Talk Mostly to Machines

Nowhere is the shift more visible than in raw network traffic. Fastly, which just launched a new “AI Firewall” along with runtime controls designed to block prompt-injection attacks, hit a milestone over July and August: pure machine-to-machine traffic crossed 50% of all traffic on its network for the first time. Between January and May, AI-driven traffic grew 6.5 times faster than traffic generated by humans. The market has noticed — Fastly shares are up 144% year-to-date and recently traded near $20.97. As machines increasingly talk to other machines rather than to people, the network layer itself becomes the choke point that has to be defended, and infrastructure providers are being pulled to the front line whether they planned for it or not.

Should investors sell immediately? Or is it worth buying Morgan Stanley?

AI Is Rewriting the Defense, Too

The change isn’t only about spending more — it’s about defending differently. Juniper Research finds that “vibe coding,” AI-generated software written without much human review, is reintroducing basic vulnerabilities into production systems at scale, a trend it expects to push global application-security spending to $13 billion by 2031. Meanwhile the sheer volume of automated attacks is overwhelming human analysts inside security operations centers. A recent SANS report found that 74% of organizations say AI is already changing the size and structure of their security teams, and some are eliminating traditional vulnerability-management teams altogether in favor of fully automated triage. The vendors that can genuinely use AI to cut through alert fatigue and handle routine threats without a human in the loop are the ones positioned for real margin expansion in the coming quarters.

The Foundation Keeps Getting Poured

None of this happens without the underlying data center buildout, which shows no sign of pausing. Cantor Fitzgerald just raised its Microsoft price target to $608, pointing to an expected $175 billion in capital expenditures for calendar year 2026. Stifel, separately, reiterated its buy rating on Oracle, citing continued progress on the company’s massive data center construction. As long as the hyperscalers keep pouring hundreds of billions into new compute, the security spending required to protect it is essentially guaranteed to follow.

What It Means

The logic here is almost mechanical: there is no scalable enterprise AI without a parallel, forced upgrade of the cybersecurity stack underneath it. That gives investors something genuinely rare right now — a growth story that keeps compounding almost independently of the rate debate dominating everything else in this market. While the rest of the tape waits on the Fed’s next move, cybersecurity spending is simply happening, invoice by invoice, whether capital is cheap or not.

Best regards,
The StocksToday.com Editorial