Akamai, Fastly and Cloudflare: How Late Entrants Compete Against a First Mover

Market entry
Written By Murray Clarke

Akamai invented the commercial CDN market and could afford to go broad. Fastly and Cloudflare arrived more than a decade later and had to pick a segment the incumbent served poorly, then own it before moving up. The public record shows both approaches working, and shows what each one asks of the leader running Japan.

I've had an unusual view of this market. A team I led ran Japan hiring for Akamai under an RPO arrangement. Our agency team supported Fastly with its early Japan hires, and more recently I worked with Fastly on the appointment of Rumi Oku as its Japan leader. Watching the three companies compete in Japan over that time has been interesting and educational.

Akamai was first, so it could go broad

Akamai came out of MIT. It was incorporated in August 1998, launched commercial service in April 1999 and named Yahoo! as a charter customer. For years there was no serious alternative for a large company that needed its content delivered quickly around the world.

A first mover with that kind of lead can sell to everyone: media, retail, banks, government, software. It can build a large enterprise sales force, sign long contracts and price for the value of being the default choice.

The cost of that position shows up later. Delivery is the business Akamai was built on, and it's now the part that's shrinking. In 2025, security revenue was $2.243 billion, up 10%, delivery revenue was $1.257 billion, down 5%, and cloud computing revenue was $708 million, up 12%. Akamai's own annual report puts the delivery decline down to lower pricing on contract renewals, after a 15% fall the year before.

Akamai has handled that well. It moved early into security and now earns more from security than from delivery. But the price pressure on renewals tells you competitors are now winning on delivery and buyers know it.

Fastly went narrow: developers and real-time content

Fastly was founded in March 2011 and didn't try to match Akamai's breadth. Its prospectus opened with the line that it was in the business of enabling developers, and its pitch was a programmable edge that engineering teams could control directly. When it priced its IPO in May 2019, the customers it named were Pinterest, The New York Times and GitHub.

That focus gave Fastly deep relationships with some of the most demanding customers on the internet. At the end of 2018, 227 enterprise customers generated 84% of Fastly's revenue. Fastly reported a record fourth quarter for 2025, with full-year revenue of $624.0 million, up 15%.

Fastly competes with Akamai head on. Its analysis of popular websites that moved from Akamai to Fastly between June 2024 and February 2025 shows a median 57% improvement in time to first byte. These are Fastly's own figures, but they show why performance-sensitive sites make the switch.

In Japan, Fastly came in through a partner. In 2015 it set up Fastly K.K., opened an office in the Shin Marunouchi building, added an Osaka point of presence and signed SoftBank as its primary sales route. Its publicly named Japanese customers include Nikkei, DeNA, which has used Fastly's Next-Gen WAF since 2018, Oak Lawn Marketing, which runs Shop Japan, and the restaurant sites Retty and Gurunavi. In my experience its strongest sector in Japan is gaming and entertainment, where traffic comes in sharp peaks around launches and live events.

Cloudflare went narrow from the bottom of the market

Cloudflare picked a different segment: the millions of websites Akamai never priced for. Its IPO prospectus describes a model built on free and self-serve plans, with security as the reason to sign up. At its September 2019 listing it served more than 20 million internet properties but had only around 75,000 paying customers.

Then it moved up. Cloudflare's 2025 revenue was $2,167.9 million, up 30%. It ended the year with 4,298 customers paying more than $100,000 a year, and those customers produced 73% of fourth-quarter revenue. The company that started with free accounts now earns most of its money from large enterprises, and it's about half Akamai's size by revenue.

The move up hasn't been smooth. In April 2023 Matthew Prince told investors that growth was limited by go-to-market performance, that larger deals were taking longer to close, and that the company was moving out more than 100 salespeople who had consistently missed expectations. Large enterprises also buy on reliability. In November 2025 Cloudflare had what it called its worst outage since 2019, with core CDN and security services failing for several hours, and in December a second outage hit about 28% of its HTTP traffic for 25 minutes. Selling to the enterprise takes a different sales team and a different standard of uptime from selling to small websites.

Its Japan entry followed the same order. Cloudflare had points of presence in Tokyo and Osaka well before it had a company here. It opened Cloudflare Japan in July 2020 with Masakazu Aoba, previously president of Citrix Systems Japan, as country head. Aoba-san said at launch that the Japan business would focus on large users. By the time Cloudflare hired a country head, the brand was already known to Japanese engineers, and his job was to turn that into enterprise contracts.

Where they meet in Japan: the public evidence

Japanese engineering teams write openly about switching CDNs, which gives a clear view of how these companies compete here.

GameWith moved its static content delivery from Akamai to Cloudflare from November 2025. The engineer who wrote it up said the main reason was cost. He also said that, for CDN features alone, he thought Akamai was the stronger product, and that Cloudflare's appeal was the wider platform around it.

The incumbent is losing on price, not on product, which matches what Akamai's annual report says about renewals. That leaves room for a late entrant that wins on performance and engineering depth, which is the ground Fastly chose from the start.

CDNetworks: the regional specialist

CDNetworks is the reminder that a regional niche on its own is hard to hold. It was founded in Korea in 2000, built a strong position in Asia, and was bought by KDDI in 2011 for $167 million. In 2017 KDDI sold it to Wangsu, China's largest CDN provider. It still operates in Japan, now as part of a much larger Chinese group, and has broadened into security and edge services.

What this means when you hire the Japan leader

The two strategies need different leaders in Japan.

A first mover hires for breadth. The Japan leader needs coverage across industries, relationships with the large system integrators and the patience to manage long contracts and renewals. The hard part is defending price when newer competitors arrive cheaper.

A late entrant hires for the segment it's trying to own. The Japan leader has to be credible with that specific group of buyers from the first meeting, because there's no brand halo to fall back on. Cloudflare picked a leader with the enterprise relationships to move from a known engineering brand into large accounts.

Fastly's choice of Rumi Oku follows the same logic. Fastly appointed her as 日本事業責任者 (Nihon jigyō sekininsha), head of the Japan business, from June 2026. She came from Google Cloud Japan, where she led the games, media and entertainment industry division, and before that she helped launch HashiCorp's Japan business. Gaming, media and entertainment companies are where Fastly is strongest in Japan. In my view Oku-san is extremely strong with exactly that customer base, which is why she was the right fit for where Fastly is today.

The stage matters as well. A leader who builds the first customers in a segment is a different hire from one who scales a known brand, and we've written about that difference in Builder, Translator or Caretaker. The CDN market shows the same pattern from the vendor's side. Know which segment you're trying to own in Japan before you choose who leads it. For a similar story in another category, see Dropbox vs. Box in Japan.

Sources

How this post was researched: AI read the companies' annual reports, IPO prospectuses, earnings releases and call transcripts, their English and Japanese press releases, Japanese technology press, and the engineering blogs of GameWith and Gurunavi, working under Murray Clarke's direction. The hiring history and the views on leadership are Murray's own, from more than 20 years of executive search in Tokyo. Murray Clarke is accountable for the content.

Murray Clarke
Next

Snowflake vs Databricks in Japan: two country-manager playbooks