Akamai Technologies Highlights $400M Cloud Run Rate, Edge AI Inference Push at Morgan Stanley TMT 2026

Key Points

• Akamai’s cloud infrastructure services reached roughly a $400 million run rate, driven by broad-based adoption across “hundreds” of customers rather than a few mega-deals, and management cites an egress-cost advantage versus hyperscalers.
• The company is scaling its Inference Cloud for latency-sensitive edge GPU inference — live in about 20 locations with potential to expand to 20–40 — and disclosed a $200 million, four-year deal that grew from a proof-of-concept to a large-cluster order.
• Security remains a major growth engine: API Security is above a $100 million run rate and growing >100%, renewal rates are in the “90s%” range, and Akamai adds roughly $200–$250 million of new security revenue annually.

Akamai Technologies NASDAQ: AKAM Chief Financial Officer Ed McGowan said the company is building on its legacy content delivery network (CDN) business by expanding further into cloud infrastructure and edge-based AI inference, describing the strategy as a “natural adjacency” across delivery, compute, and security.

Speaking at Morgan Stanley’s TMT 2026 event in a conversation with analyst Sanjit Singh, McGowan outlined how Akamai’s evolution has moved from its original delivery networking roots to a sizable security business and, more recently, to cloud infrastructure services (CIS) and GPU-based inference offerings. He framed the current phase as driven by customer demand and enabled by Akamai’s existing distributed platform and operational scale.

From delivery to security and enterprise compute

McGowan said Akamai’s security business grew out of the company’s role in delivering and protecting customer websites, citing the logic of combining performance and protection at the edge. He said the company has added roughly $200 million to $250 million of new security revenue annually for about a decade, supported by internal innovation and acquisitions. He also pointed to API Security—referencing the company’s Noname acquisition—as an area he believes is positioned for future application architectures with increasing machine-to-machine data flows.

On the compute side, McGowan said customers increasingly wanted more control than functions-as-a-service could provide, including the ability to run proprietary code with “root access.” He cited live streaming as an example of a use case where customers wanted custom code running closer to end users to improve performance. Akamai’s acquisition of Linode, he said, helped the company move from serving small and mid-sized users toward an “enterprise-grade system,” including capabilities such as managed container services.

McGowan also said Akamai’s internal needs played a role in the push into compute. He described the company as a major consumer of third-party cloud services due to the scale of data processed in security operations, and said Akamai has been moving workloads off public clouds onto its own infrastructure, describing it as a source of cost savings and proof that the platform can support broader customer demand.

Cloud infrastructure growth driven by breadth, not concentration

McGowan said Akamai’s CIS business has reached approximately a $400 million run rate and that growth has surprised him in one key way: customer concentration has been lower than he expected. Rather than being driven by a small number of “mega deals,” he said the business is composed of “hundreds and hundreds of customers,” ranging from smaller spenders to customers spending up to around $2 million per month.

He added that large deals are beginning to ramp and could become more material over the next year or so, but emphasized that current momentum has come from broad-based adoption and a growing partner ecosystem. Among the faster-growing use cases he cited was observability, where customers use partner tools to ingest data, store it, run compute, and build dashboards.

Asked why customers choose Linode versus hyperscalers, McGowan argued that cloud adoption is not “zero-sum,” with many companies pursuing multi-cloud strategies. He said differentiation can come from performance, latency, economics, and the ability to meet certain standards. He also noted that the major hyperscalers are customers of Akamai, using the company for things including video delivery, API management, and advertising decisioning.

On economics, McGowan highlighted customer concerns about egress fees in hyperscaler environments. He said Akamai can avoid charging for certain egress fees because it operates one of the largest backbones in the world and delivers “hundreds of terabits per second,” allowing it to compete on total cost. He did not provide a specific percentage or range for how much cheaper Akamai can be relative to hyperscalers, but described egress as a meaningful source of “hidden costs” for customers.

Inference Cloud: why Akamai expects edge deployments to matter

McGowan said Akamai’s Inference Cloud, announced in the fall, reflects growing customer demand for distributed GPU compute to support latency-sensitive inference. He described the next generation of applications—including more “agentic” workflows such as virtual travel assistants—as compute-intensive and increasingly sensitive to latency in ways similar to earlier e-commerce performance requirements. He also cited use cases such as robotics and autonomous driving as categories that require extremely low latency and benefit from compute being close to where activity occurs.

He said Akamai’s work with NVIDIA helped validate performance and economics for relatively small deployments, and that the company is seeing inbound interest from AI startups seeking GPU rentals as well as customers wanting GPU clusters for research, post-training, robotics, and real-time advertising decisioning.

$200 million deal, expansion plans, and unit economics

McGowan discussed a disclosed $200 million, four-year deal with a major tech customer, describing it as an existing Akamai customer that approached the company after the Inference Cloud announcement. He said the engagement began with a proof of concept in one location before scaling to a large cluster order, and that contract cycles from proof-of-concept to signing can take roughly two to six months. He added that revenue recognition can lag contract signing when additional data center space must be brought online.

On footprint, McGowan said Akamai currently supports Inference Cloud in about 20 locations, using existing facilities that already had the needed cooling and power to add GPU racks. Over time, he said expansion could involve a mix of deploying into existing sites with capacity and adding space or new sites for large, location-specific clusters. He suggested the GPU footprint could grow from 20 locations to “somewhere 20–40” based on demand, while also noting that running GPUs across thousands of Akamai points of presence would not likely be economical.

McGowan also walked through a framework for unit economics, stating that a 1,000-GPU deployment requires about a megawatt of power. He estimated annual power costs in the U.S. at roughly $2 million to $4 million (and higher in California), and referenced an estimated $12 million to $16 million for hardware and supporting infrastructure—using $20 million as a conservative upper-end assumption. He pointed to a rental market rate of about $2.50 per hour at launch, and said that even with discounting, he believes the economics can support strong margins, describing potential gross margins in the “70%-ish” range and operating margins north of 30% under his illustrative assumptions.

Margins, delivery pricing, and security product momentum

McGowan acknowledged recent margin pressure tied to higher capital investment and costs associated with co-location and depreciation. He said co-location costs can be elevated when capacity is built ahead of revenue and also due to non-cash accounting effects from multi-year leasing commitments that require straight-line expense recognition. He characterized this as part of a broader trade-off between growth and margins, with operating leverage expected over time in go-to-market, engineering, and operations due to shared infrastructure and teams across delivery, security, and compute. He said Akamai has moved about 1,000 engineers from CDN work into compute-related work given the overlap in operating a distributed platform.

In delivery, McGowan said conditions have improved, including a better pricing environment and fewer competitors. He also noted that customer rebalancing following Akamai’s acquisition of Edgio assets contributed to delivery revenue declines during the year, as multi-CDN customers shifted portions of traffic to other providers over time. McGowan said Akamai is now taking a tougher stance on pricing—seeking increases on renewals and resisting steep price declines—adding that the company is raising price “for the first time in our history,” citing higher costs including co-location and memory-driven capital spending.

On security, he highlighted areas he described as fast-growing:

• API Security: McGowan said it exited at over a $100 million run rate and was growing at over 100%, with penetration in the install base below 10%.
• Guardicore: He said penetration is also below 10%, with much of the business coming from new customers and expansion driven by adding more agents across customer networks.
• License renewals: He described renewal rates as in the “90x%” range, with term licenses often expanding over time.

McGowan closed by reiterating confidence in continued growth across Akamai’s security and compute initiatives, with delivery trends improving as pricing discipline tightens and the company expands its infrastructure to meet customer demand.

About Akamai Technologies NASDAQ: AKAM

Akamai Technologies, Inc is a leading provider of content delivery network (CDN) services and cloud security solutions designed to optimize and safeguard digital experiences. Leveraging a globally distributed platform, the company accelerates web and mobile content delivery for enterprises, media companies, e-commerce platforms and government agencies. Its edge computing architecture brings processing power closer to end users, reducing latency and improving application performance across geographies.

The company's core offerings include content acceleration, web and mobile performance optimization, media delivery, and a suite of cybersecurity solutions that protect against DDoS attacks, application-layer threats and bot-driven fraud.

This instant news alert was generated by narrative science technology and financial data from MarketBeat in order to provide readers with the fastest reporting and unbiased coverage. Please send any questions or comments about this story to contact@marketbeat.com.