Fastly: Riding the Agentic AI Wave

Fastly is presented as a potential beneficiary of agentic AI because its edge-computing and security services can support frequent, low-latency agent queries. In 738 tests, Meta’s Muse agents sent roughly one-third of their traffic through Fastly, though the article says Meta has not publicly confirmed using its technology. Zacks estimates Fastly earnings growth of 71.43% next quarter and 307.69% for full-year 2026; the company beat consensus estimates by an average of 92.26% over the past four quarters, and Oppenheimer upgraded the stock to Outperform.

Analysis

The investable question is not whether agents create more requests, but whether Fastly can convert that activity into profitable, billable traffic. More frequent calls can lift usage, yet much agent work still depends on centralized model inference; edge routing, caching, security, and lightweight execution are complements to that compute, not substitutes for it. If requests are dynamic or uncachable, higher volume may bring bandwidth and operating costs without attractive incremental margins.

Treat the reported Meta traffic test as a lead, not commercial proof: IP attribution across a limited set of test sites does not establish a Meta contract, broad deployment, or material revenue. The bullish thesis needs customer-level evidence and improving revenue quality. Agent traffic also creates a product tension: security tools that block unwanted bots may need to distinguish authorized agents from scraping, and restrictive customer policies could limit monetization.

Over the next 1–3 months, the key catalysts are earnings commentary and measurable evidence of AI-related usage converting into revenue, gross-margin progression, and security/compute attach. The 6–18 month opportunity depends on whether edge workloads become a repeatable, differentiated offering rather than a generic CDN feature. Cloudflare and Akamai are relevant competitive checks; hyperscalers may also build or bundle parts of the stack. The contrarian risk is that the agent narrative gets capitalized before Fastly demonstrates economics. Validate whether cited EPS growth reflects a low comparison base or adjusted metrics, and compare it with revenue and cash generation.

Key Decisions for Investors

• Do not chase a narrative-driven breakout on the article alone. Consider a small, risk-budgeted FSLY position only if the technical breakout holds; define the exit at a failed breakout/base, since no price levels are supplied.
• Use the next earnings report as the 1–3 month confirmation point: look for explicit AI/agent customer usage translating into revenue, improving gross-margin economics, and management disclosure that can be independently checked. Without that evidence, keep FSLY on watch rather than adding.
• Track Cloudflare and Akamai disclosures alongside Fastly. If competitors report similar AI-traffic demand with stronger monetization, the category may be real while Fastly’s differentiation is not; that would weaken a standalone FSLY long.
• Falsify the bullish thesis if traffic growth fails to improve revenue or margins, management does not substantiate AI monetization, or Fastly loses the breakout/base. Avoid treating the reported Meta test as evidence of a material paid deployment.