Fastly, Inc. Class A Common Stock · Moat analysis

Report previewFSLY.USReport date Last updated
← Back to reports
Report contents
01

Industry landscape

Where the company competes: a subscale vendor in a crowded, global edge market

Fastly runs the layer between a customer's servers and its end users. It caches and serves video, images and API responses from machines close to the viewer, filters the traffic that hits those machines, and now sells a small amount of compute on the same hardware. The precise arena is performance-sensitive content delivery and edge security for large enterprises: streaming and media companies, e-commerce, SaaS, financial services and, increasingly, AI operators. That is narrower than "cloud" and narrower than "security." Fastly does not sell general-purpose compute or a broad application-security suite to midmarket buyers.

The revenue mix shows how lopsided that arena still is. Figures below are for the quarter management discussed at Citi's 2026 Global TMT Conference on September 9, 2026, as reported by Investing.com.

Product lineRevenueShareYear-over-year
Delivery$133.6 million73%+17%
Security$42.1 million23%+43%
Other and compute$7.3 million4%+69%
Total$183.0 million100%+23.3%

Delivery pays the bills and grows slowest. Security and compute are small but compounding at roughly twice the corporate rate, which is why the mix question matters more than the headline growth rate. Total revenue grew 23.3% while the non-top-10 cohort of 624 large customers grew 12%, so the acceleration came from the biggest accounts, not from broad-based demand.

The value chain explains the margin structure. Upstream, Fastly leases colocation space and power, buys servers, and pays for transit and peering with internet service providers; bandwidth is its largest variable cost and it resells that capacity per gigabyte. Downstream, customers pay for delivered traffic, security subscriptions, or compute time. Because the 166 points of presence are fixed assets, gross margin depends on utilization. The company reported record gross margin of 65.8% and capex at 10% to 12% of revenue for the quarter, with absolute capex up 40% to 45%, so it is still adding capacity ahead of the traffic that will fill it.

Competition is global in engineering terms and concentrated in sales terms. Latency is measured across borders, and peering relationships in Frankfurt or Singapore matter to a customer headquartered in San Francisco. But the buying decision sits with a few hundred large enterprises, mostly in North America and Europe, and Fastly's own disclosure that its top 10 customers generate 37% of revenue says the addressable arena is narrower than the network map implies. Scale helps globally; it does not confer dominance in any one region.

Market structure is the weak part of the story. Delivery has at least six credible global vendors, plus regional CDNs and telecom carriers selling caching as an add-on.

VendorWhere it overlaps with FastlyBasis of competition
AkamaiDelivery, WAF, bot managementIncumbent scale; bundling delivery with security
CloudflareDelivery, WAF, bot management, edge computeDeveloper adoption; one platform, many products
Amazon CloudFrontDelivery, WAF, edge computeBundled with AWS; can price delivery as an enticement
Google Cloud CDN, Azure Front DoorDelivery, WAFBundled with their clouds
FastlyDelivery, security, computePerformance niche; unified stack on one network

Public market-share estimates for CDN are not comparable across sources: some count only delivery revenue, others fold in security and edge compute, and no vendor publishes a defensible number. So share cannot settle the concentration question here.

The practical test is more useful. If the leaders cannot be counted on one hand, entry barriers are usually weak, and by that standard pure delivery qualifies: commodity caching is easy to enter, prices fell mid-to-high single digits year over year, and management itself credits "industry rationalization and the exit of weaker competitors" for the recent stability in pricing. What is harder to replicate is the combination Fastly now sells: a single anycast network running delivery, five security products and compute on identical infrastructure. That raises barriers to medium for the full platform, and low for the delivery piece alone. The other constraint is bundling. AWS, Google and Microsoft can treat delivery as a customer-acquisition cost, which caps what a standalone vendor can charge regardless of technical merit.

So the arena is a crowded, global, moderately concentrated services market where Fastly holds a small share and competes on performance and product breadth. The open question is whether security and compute grow fast enough to change which budget Fastly is selling against. If security keeps compounding above 40% and net revenue retention holds near the reported 117%, the company is increasingly a security vendor with a delivery attachment. If that growth cools, the business reverts to per-gigabyte pricing in a market with too many sellers.

02

Competitive advantage

Does the company have a moat? Weak: the returns test fails and the share test cannot be run

Moat rating: Weak. Fastly has not earned its cost of capital in any year as a public company. It reported an operating loss in each fiscal year from 2021 through 2025, and after-tax ROIC was negative throughout, so the profitability side of the moat test fails on the numbers rather than on a judgment call. The concentration side cannot be measured cleanly, because neither Fastly nor its rivals publish a comparable CDN share series. What can be measured points to a subscale participant: revenue of $0.62 billion in FY2025 against much larger competitors, and a customer base in which ten accounts supply 37% of revenue. Fastly listed in April 2019; the five fiscal years through 2025, plus the first three quarters of 2026, are the comparison window.

Market position: outside the top tier, and the share data is not public

Market position: Weak. Fastly sells content delivery, security and a small compute line to large enterprises. It discloses no unit or revenue share, and no third party publishes a series that covers Fastly and its rivals on a like-for-like basis. On revenue alone, an estimate based on relative scale puts Fastly outside the top four globally: Akamai and Cloudflare each report several times Fastly's revenue, and AWS, Google and Microsoft sell delivery as an attachment to compute they already own. In Fastly's strongest niche, security for large media and e-commerce properties, it is at best a strong number three.

High gross margin does not rescue the position. Fastly's gross margin of 57.1% in FY2025, rising to roughly 63% in the 2026 quarters, looks like pricing power in a business that resells bandwidth. Part of it is mix rather than price: security is priced as software, and the largest accounts buy enough volume to negotiate hard on delivery.

Stability of position: growth from a small base is not protection

Stability: Moderate. Revenue rose from $0.35 billion in FY2021 to $0.62 billion in FY2025, and grew 23.3% year over year in the September 2026 quarter. A company without a moat can still grow, and this has that shape. The acceleration came from the top-ten cohort while the 624 large accounts outside it grew 12%, which is what winning custom engineering work for a handful of big customers looks like.

The framework's rule of thumb cuts against Fastly here: average share that moves more than 5 percentage points over five years suggests entry barriers are absent, while a move under 2 points suggests very strong ones. Fastly's share is too small to measure at that precision and is probably rising, which is the opposite of the stability the test rewards. A share gainer in a market with six credible global sellers is more likely to be out-executing rivals on specific workloads than to be protected by a barrier to entry.

Profitability: five years of losses, narrowing quickly
Fiscal yearRevenue (USD bn)Gross marginOperating marginNet marginROE
20210.3552.9%-63%-63%-21.5%
20220.4348.5%-58%-44%-19.4%
20230.5152.6%-39%-26%-13.8%
20240.5454.4%-32%-29%-16.3%
20250.6257.1%-19%-20%-12.8%

Profitability: Weak. Average ROE across those five years was about -17%. ROIC is the better test and cannot be computed as a meaningful positive figure: with operating income negative every year, after-tax returns ran below zero on invested capital of roughly $1.3 billion (equity plus debt). The framework treats a 15%-25% average after-tax ROIC as the signature of a protected franchise and 6%-8% as evidence of none. Fastly sits below zero, a different category altogether.

The trend is the one genuine positive. Net margin improved from -63% in FY2021 to -20% in FY2025 and to -5.9% in the September 2026 quarter, when operating income was about -$10 million. Operating cash flow reached $90 million in FY2025 and was positive in each of the first three quarters of 2026. That is progress toward breakeven, not evidence of excess returns.

Margins: Moderate. Gross margin rose from 52.9% to 57.1% between FY2021 and FY2025 and reached roughly 63% in the 2026 quarters, above what per-gigabyte delivery alone would earn. Net margin has never been positive, so the company has not yet shown it can convert that gross margin into profit at its current size.

What the two tests add up to

Moat rating: Weak. The position is subscale against rivals that bundle, and the return record is negative. When profitability falls short of Fairly Strong, the overall rating should not exceed Moderate; combined with a small and probably rising share, Weak is the defensible label. It is not None because gross margin above 60% and the retention management reports say customers get something from Fastly they cannot get elsewhere, which the next section examines. Two years of positive operating margin with ROIC in double digits, alongside security revenue still compounding above 40%, would reopen the question. Nothing in the record so far does.

Where the advantage comes from and how long it can last: switching costs without scale

Fastly's advantage is real but narrow, and it sits in one place: the switching cost of code customers write into Fastly's edge. The supply side is replicable capital, scale runs against the company, and durability turns on whether security grows large enough to change what customers are buying before cloud bundling compresses delivery further.

Demand side: real switching costs, confined to custom code

Moderate. Start with the mechanism. A Fastly customer does not just point a DNS record at the network. Engineers configure caching rules, routing logic, header manipulation, authentication and image transforms that run on Fastly's servers nearest the end user, and the edge compute product extends that with custom code. The configuration is business logic living on Fastly's infrastructure. Replacing it means porting, retesting latency and correctness, and recertifying with the customer's own teams. A migration consumes months of engineering time even when the new vendor quotes a lower price per gigabyte. That is genuine captivity, and it explains why revenue from existing accounts grows even in quarters when Fastly wins few new logos.

Three things keep the rating at Moderate.

  • The underlying workload is standardized. Delivery is measured in benchmarks: time to first byte, cache-hit ratio, p95 latency. Buyers can compare vendors cheaply before signing, so search costs stay low even where switching costs are high.
  • Multi-CDN is normal practice. Large media and e-commerce companies run two or three networks at once to manage risk and keep vendors honest on price. With ten customers at 37% of revenue, the accounts that matter most have both the engineers and the leverage to dual-source. Captivity that can be split across vendors binds only at the margin.
  • The code is portable in principle. Fastly's configuration language is proprietary, but the functions it performs exist in every rival's product, and challengers supply migration help free to win the account. The barrier is time and risk, not inability to move.

Price is the test that settles it. A captive base shows pricing power, and in delivery Fastly's prices have kept falling at a mid-to-high single-digit rate. A customer that concedes price every year is buying a service, not locked into one.

Supply side: a leased network rivals can replicate

Weak. A supply advantage usually means proprietary technology, a scarce input, or a cost structure others cannot match. Fastly has none of the three in defensible form.

It owns little of the physical layer. It leases colocation space and power, buys commodity servers, and pays for transit and peering. The 166 points of presence are capital, not a scarce resource; no fiber route or right of way is exclusive to Fastly. The architecture is standard too. Anycast routing, caching and TLS termination are well understood, and every competitor runs a version of the same design. What Fastly has is operational craft: tuning cache-hit ratios, holding peering in specific metros, keeping performance steady through a traffic spike. That know-how sits with engineers, depreciates when they leave and spreads when rivals hire them. No patent portfolio in the disclosures would block a determined entrant, and the foundational caching patents in this industry expired years ago.

The craft does show up in gross margin, up from 52.9% in FY2021 to 57.1% in FY2025 and about 63% in the 2026 quarters, helped by security mix and better utilization of capacity already built. That is worth real money. It is also the kind of edge a larger network can match by spreading the same fixed cost over more traffic.

Scale: high fixed costs, unfavorable relative share

Weak. An edge network is fixed-cost heavy: points of presence, servers, peering, round-the-clock operations, and R&D across caching, WAF, bot management and compute. Fastly ran about $480 million of operating expense against $0.62 billion of revenue in FY2025, and gross profit of $0.36 billion did not cover it.

Minimum efficient scale is real in this business. A vendor needs coverage in Frankfurt, São Paulo and Singapore before it can pass a large enterprise's latency tests, and that entry cost is meaningful. The tell is that Fastly cleared the bar and still does not earn its cost of capital. Scale sufficient to compete is not scale sufficient to protect.

Relative share then moves the wrong way. The framework cares about share against competitors in the specific market, not absolute size. Akamai and Cloudflare carry more traffic over which to spread the same peering and R&D. AWS, Google and Microsoft can price delivery as a customer acquisition cost inside a larger cloud bill, which caps what a standalone vendor can charge regardless of latency. Fastly's defensible scale is confined to security, 23% of revenue, where one product line amortizes its R&D across the same few hundred accounts that already buy delivery.

Durability: bundling, not a rival CDN, is the threat

The substitute that matters is not another content delivery network. It is the cloud edge: caching and functions sold by the vendor that already hosts the application and priced into a committed-spend contract. For a customer running workloads in AWS or Azure, shifting standard delivery to the in-house CDN is a procurement decision rather than an engineering project, and it retires a vendor relationship. For most static and video traffic, that substitute is good enough.

Two developments would strengthen the case, and one would break it. Strengthening: security grows from 23% toward half of revenue at 40%-plus growth, moving the buying conversation from bandwidth price to risk reduction, where switching costs are higher and hyperscalers are less dominant. Also strengthening: net revenue retention holds above 115% through a cycle of falling delivery prices, which would show that code-level lock-in survives commercial pressure. Breaking: a hyperscaler folds edge delivery and WAF into committed cloud spend at a discount Fastly cannot match, or AI inference traffic consolidates onto the networks of the companies that own the GPU capacity.

Regulatory barriers do not apply. Fastly holds no license a rival needs, and no regulator sets its prices or limits entry.

What is left of the advantage

The strongest dimension is demand-side captivity at the code layer, and it earns a Moderate at best. The weakest is supply side, where the network is leased, the design is standard and the know-how walks out the door. Scale works against the company rather than for it. Positive free cash flow and a $1.0 billion equity base give Fastly room to keep building; they do not keep rivals out. Until security changes what customers are actually buying, returns will keep depending on execution in a market with too many sellers, and the record so far shows execution improving without a structural barrier forming around it.

Generated by DeepVal AI. Data and models may contain errors or be outdated. For research only; not personalized investment advice.

Continue the comparison

Continue reading the full report

  1. Comparison and summaryHow Fastly, Inc. Class A Common Stock compares with its closest rivals, and our moat rating

Free to sign up · 20 research points · unlocking this report uses 10

20 points cover 2 full reports. No card needed.

Research reflects its generation date. Preview the first two modules.

More on Fastly, Inc. Class A Common Stock