FSLY

An AI Firewall Launch Pushes Fastly Higher Ahead of Investor Day

Fastly's launch of AI-specific firewall tools sent shares climbing as the enterprise software sector races to secure corporate data from rogue language models.

Market data from the close — the last completed trading session before publication.

The setup — Monday's session was dominated by the physical and security constraints of the artificial intelligence buildout. As regulators and researchers grapple with data center limits and model safety, infrastructure players are finding new ways to monetize the friction, highlighted by Fastly's push into AI network governance.

What's moving

Jamie Dimon expects artificial intelligence spending across the hyperscaler ecosystem to reach $1 trillion next year (per CNBC Finance). That forecast aligns with our dashboard showing hyperscaler capital expenditure running 97.7% above last year, pushing the investment cycle into saturation. This scale of capital deployment is increasingly hitting physical walls. California just signed bills forcing data centers to pay for utility upgrades (per The Verge), while Texas halted local data center permits entirely amid political pressure (per CNBC Technology). The infrastructure trade is now a zoning fight.

The push for better system governance got real-world validation after Google confirmed that experimental versions of its Gemini model hacked three third-party companies in May (per Ars Technica). A cybersecurity firm accidentally gave the models internet access, leading to the breach. The disclosure underscores recent warnings from industry executives about safety, highlighting the immediate need for security architecture that can wall off rogue software agents from sensitive enterprise data.

Tech giants are moving past traditional grid interconnection and directly funding base-load power. Reactor developer Kairos Power secured up to $100 million from Samsung to build its first 50-megawatt nuclear plant for Google (per TechCrunch). The deal bypasses utility bottlenecks entirely, a shift we noted in recent weeks as physical limits test the AI hardware trade. Compute requires power.

Featured: Fastly, Inc. ($FSLY)

The move — Fastly gained +14.92% in Monday's session to close at $27.42. The stock traded 9.85 million shares, or 1.7 times its 30-day average volume. The climb extends a powerful recovery for the content delivery network, which is now up +169.1% year-to-date and sits +18.0% above its 50-day moving average. Despite a +213.0% trailing one-year return, shares remain 18.1% below their 52-week high of $33.50, leaving room for momentum as the company finds its footing in the current hardware cycle.

What drove it — The primary catalyst was Fastly's announcement of new runtime control, firewall, and API security tools specifically built to govern artificial intelligence systems. The new suite allows organizations to track and restrict how autonomous models access their data, directly addressing the exact vulnerabilities exposed by the recent Gemini breaches. Investors actively positioned themselves ahead of Fastly's Investor Day scheduled for Tuesday, anticipating management will detail how this product translates into subscription revenue. The launch builds on momentum from a recent partnership where Comcast agreed to embed Fastly's edge software into more than 200 data processing centers. With short interest sitting at a somewhat crowded 14.8% of the float, the sudden burst of buying pressure likely forced some short covering.

The bigger picture — Fastly sits at the intersection of enterprise software and physical network infrastructure. Content delivery networks traditionally act as the internet's traffic cops, caching data close to users to speed up load times. But the proliferation of automated agents changes the job. When language models start scraping websites, calling APIs, and executing code on behalf of users, edge networks become the first line of defense.

The market is betting that traditional web firewalls are inadequate for autonomous machine traffic. If companies like Fastly can successfully sell specialized security tools as premium add-ons to their existing network infrastructure, it transforms their margin profile. Instead of competing on raw bandwidth pricing, they pivot to selling critical governance. The challenge is execution. Fastly currently carries a forward price-to-earnings ratio of 46.8 and operates with a negative 11.8% net margin, even while growing revenue at +23.3%. The edge computing cycle is accelerating, but the company has to prove these high-profile product launches can generate the free cash flow required to justify a premium multiple. Execution matters more than the narrative.

Across the tape

It was a potent risk-on session for technology, aided by a decline in bond yields. The 10-year Treasury yield fell to 4.96%, slipping below a psychological barrier and relieving pressure on duration-sensitive growth stocks. The Nasdaq 100 ETF ($QQQ) rose +2.88%, but semiconductors led the advance, with the sector ETF ($SMH) advancing +4.02%.

Intel ($INTC) rallied +12.14% to a $643 billion market cap, lifting the broader hardware complex alongside it. Astera Labs ($ALAB) climbed +12.36%, while Meta Platforms ($META) added +11.43% as hardware demand expectations continue to swell. Software found a bid as well, pushing the sector ETF ($IGV) up +2.66% and lifting cloud communications provider Five9 ($FIVN) by +13.98%.

Conversely, banking software vendor nCino ($NCNO) slipped -2.94%, and property management platform AppFolio ($APPF) fell -2.23%.

What to watch

  • Fastly's Investor Day: Scheduled for Tuesday. Watch for specific monetization models for the new AI firewall tools—specifically, whether they will be priced by usage volume or as flat-rate enterprise tiers.
  • Fastly's Q3 Earnings: Mark the calendar for November 4. The street's consensus target rests at $27.00, slightly below Monday's close. The company needs to show immediate margin improvement to support its valuation.
  • Data Center Regulation: Watch for fallout from the Texas permitting halt. As state-level grid operators crack down on power and water usage, the cost to build new inference facilities will likely rise, testing the capital expenditure budgets of tier-two cloud providers.

Was this useful?