1998Susumu Fujita founds CyberAgent in Tokyo, aged 24
1998CyberClick, a click-guaranteed ad product, launches
2000Lists on the TSE Mothers market
2001The Murakami Fund files a shareholder proposal
2003“Lifetime employment” declaration
2004CAJJ program introduced; the Ameba blog opens
In March 1998 Susumu Fujita, twenty-four, left the staffing firm Intelligence and set up CyberAgent in Minato, Tokyo. It began with no product of its own: its first income came from acting as a sales agent for WebMoney, earning purely on its selling. In July 1998 it launched CyberClick, a click-guaranteed banner-and-text ad product, and turned itself from a sales agent into an internet-advertising company. Crucially, it outsourced all system development to On the Edge (the future Livedoor) under a five-year exclusive deal paying 10% of sales as royalties, keeping its own people on selling and client work. By January 2000 it had lined up 4,728 ad-carrying media — selling power, not a product, carried the early margins.
In March 2000 CyberAgent listed on the Tokyo Stock Exchange’s Mothers market, and its market value briefly touched $3.6B (¥390bn). It put the $192.1M (¥21bn) it raised into an investment-incubation business, backing internet ventures — just as the dot-com bubble burst and those holdings collapsed in value. Raising over ¥20 billion two years after founding was extraordinary for a company its age, and the write-downs soon ate into the profits of the core ad-agency business. In 2001 the Murakami Fund, spotting a company whose cash exceeded its market value, launched a shareholder proposal — an early collision between founder control and outside capital.
Fujita chose not to shrink the business and hand back the cash, but to absorb the losses with the ad agency’s profits and press on. In 2003 he issued a “lifetime employment” declaration to hold down turnover, and by the year ended September 2004 the group swung back to profit and formally opened the investment-incubation segment. In May 2004 it introduced the CAJJ program, splitting new ventures into subsidiaries graded J3/J2/J1 on rolling gross profit and — unusually — disclosing the withdrawal thresholds to outside investors. In September 2004 it launched the Ameba blog. The template of many disclosed-criteria subsidiaries, and of a sales-only startup becoming a company that owns its own media, was set here.
2014Ameba restructured; headcount halved from 1,600 to 800
In July 2005 CyberAgent set up an Ameba division and pushed into media in earnest. In 2006 it began hiring engineers — a heavy pivot for a company whose centre of gravity had been its sales force — because once it decided to own media, it had to own the technology behind it. That in-house engineering later became the seedbed for the Cygames games business and the ad-tech that powers its ad delivery. Under the cover of the ad agency’s profits, Ameba was allowed to run at a loss until it turned profitable in June 2010, giving the group a revenue source that did not depend on selling ads.
In May 2011 Fujita declared a “smartphone shift,” reallocating development from PC to mobile ahead of most of the industry, and in 2013 sold off CyberAgent FX to concentrate resources. In September 2014 he restructured the Ameba business, killing low-margin services and halving its headcount from 1,600 to 800 — a decision that reached straight into the organisation’s structure three years after the shift he had declared. Most of those people were redeployed into the group’s games and ad-tech units rather than let go, keeping the know-how in-house.
The smartphone shift let mobile games grow into a segment second only to advertising. The customer base built in the ad agency and the development capacity built for the shift combined to spawn the games business; the group’s model of semi-independent games subsidiaries kept studios autonomous while sharing ad and platform know-how. A one-legged advertising company had become a three-domain group — advertising, games and media.
2015AbemaTV founded as a joint venture with TV Asahi
2018Capital tie-up with FC Machida Zelvia — into J-League ownership
2019Approval for Fujita’s reappointment falls to 57.56%
2021Record profit on the game Uma Musume
2022ABEMA streams all 64 FIFA World Cup matches free
In April 2015 CyberAgent set up AbemaTV as a joint venture with TV Asahi and entered internet television. Cumulative investment passed $761.2M (¥100bn) and AbemaTV’s negative net worth reached $845.7M (¥111bn), but the portfolio held: the profits of advertising and games absorbed the media losses. This was the same structure as the founding-era arrangement in which the ad agency’s profit tolerated the incubation losses — the same shape, an order of magnitude larger. The TV Asahi tie-up joined terrestrial programming know-how to streaming operations.
The magnitudes of the cycle show in the numbers. AbemaTV’s negative net worth swelled from ¥66.5 billion in FY2019 to $845.7M (¥111bn) in FY2022, and FY2019 carried a net loss of $178M (¥19bn). Against that, in FY2021 the games segment earned ¥96.4 billion and internet advertising ¥22.5 billion while media stayed at negative ¥15.1 billion — two profit engines carrying one loss-maker, quantified. In 2018 the group brought in Zelvia, operator of the J-League club FC Machida Zelvia, for $10.4M (¥1bn), and in March 2019 moved its head office to Abema Towers.
The founder’s control met outside discipline at the December 2019 AGM, where the approval rate for Fujita’s reappointment fell to 57.56% after the proxy adviser ISS opposed on the thinness of outside directors. Meanwhile the earnings swung on a single hit: in the year to September 2021 CyberAgent reached a record $6.1B (¥666bn) in sales and $950.1M (¥104bn) in operating profit on the strength of the game Uma Musume, yet on an unconsolidated basis it booked a ¥90 billion provision against loans to AbemaTV and fell to a $628.5M (¥69bn) net loss. In 2022 it took the domestic rights to the FIFA World Cup and streamed all 64 matches free on ABEMA, spending game profits to buy awareness rather than subscribers — a bet with no fixed payback date.
2023Succession plan announced; Nelke Planning joins the group
2024Anime & IP division created; Nitroplus joins the group
2025Anime studio CA Soa established; leadership handover brought forward
In March 2023 CyberAgent announced a succession plan built around promoting a successor from within, and the board began cultivating candidates. Rather than clinging on, Fujita — prompted by an internal document that imagined his own future self — chose to move early, grooming 16 candidates aged in their forties or younger over several years and putting the tacit knowledge of running the company into a written handover. The chain runs back to the 2019 shareholder revolt that forced governance reform, whose committee became the vessel for this plan: it was not sudden, but placed on top of institution and preparation.
Even so, weaning the group off founder-dependence is not done. The handover, first set for 2026, was pulled forward to 2025, and Fujita stays on as chairman with continued CEO-level involvement under a two-representative structure. The heaviest recent move is to pull the makers of intellectual property inside the company by acquisition: it added the stage producer Nelke Planning in June 2023 and the content producer Nitroplus in July 2024, created an Anime & IP division in February 2024, and set up the anime studio CA Soa in January 2025. A company that at its founding outsourced its ad-delivery system to On the Edge is moving from buying in the content it supplies to games and ABEMA toward generating, in-house, works whose rights it holds.
The bubble’s cash bought a long horizon for investment
What this listing left behind sorts into two things. One was a painful education as a manager facing the capital market: the fall from a $3.6B (¥390bn) market value to one-eighth of the share price brought home to the young Susumu Fujita his responsibility to shareholders, and the way a public company’s management is caught between the market’s expectations and reality. The other, paradoxically, was the $192.1M (¥21bn) it could raise precisely because it was the peak of the bubble. This vast cushion, raised by a loss-making company, weighed on it at once as write-downs on the investment-incubation business, yet depending on how it was used it could also become a long-term source of investment capital.
In fact this cushion set the character of the company that followed. The 2001 fight, in which the Murakami Fund closed in by exploiting the distortion of a company holding so much cash, is the symbol of it. On the other side, the 2000 turn of policy from agency toward owned media bore fruit in Ameba in 2004, and the bubble-era money became the foundation for long-term investment that raised media while tolerating losses. The listing’s frenzy and crash looked in the short run like a painful failure, yet they left the company both financial capacity and the idea of “owning our own media.” One can read this experience of tasting heaven and hell at once as the origin of its later investment judgement and financial management.
Cash as a weakness, and the long-term investment he held to
What this contest exposed was the irony that the financial cushion won at the peak of the bubble could turn straight into a weakness threatening the company’s survival. When market value falls below cash on hand, a company becomes a target for a takeover after that cash. The Murakami Fund’s logic — return idle capital to shareholders — was, from the capital market’s side, a kind of legitimate argument. Even so, Susumu Fujita did not take the road of returning the raised money as demanded and shrinking the business. Rather than liquidate and hand out cash, he bet on raising the next business — media — even while carrying losses. Asked to choose between short-term shareholder returns and long-term business investment, Fujita answered clearly for the latter.
That what saved the crisis was not a hostile defence measure but investment from Hiroshi Mikitani of Rakuten — a fellow entrepreneur — also captures the character of this episode well. Fujita did not forget the debt in later years: in 2023 he invested $71.2M (¥10bn) in Rakuten. And the company’s structure — holding no stable shareholders, its control liable to be swayed by outside large holders — did not vanish in 2001. The tension with shareholders over holding ratios and board independence surfaced again in 2019, when the approval rate for the president’s reappointment plunged to 57.56%. The first confrontation, provoked by an excess of cash, was the starting point of a question that would continue: how a founder faces the capital market.
The heart of this decision was that it changed the very structure of how the company earned. Ad-agency work is a business of selling other companies’ ad space and one’s own sales muscle, and growth runs roughly in proportion to added headcount. Ameba, by contrast, owns its own media and sells the page views it gathers as ad inventory. Moving to a structure where revenue grows as readers grow, without adding people, meant — in exchange for the resolve to carry years of loss-making up-front investment — gaining a business with leverage. That Susumu Fujita, while the top executive, directly ran a single business was a stone laid to see this shift genuinely through.
The experience of becoming an owner of media defined the company’s later years beyond the success or failure of any one business. The practice of continuing long-term investment while covering losses with the core business’s profits, and the notion of setting a user metric like monthly page views at the centre of management, took shape here at Ameba. The idea of “owning our own media,” born amid the 2000 share-price crash, took the form of a blog here, and would be handed down — an order of magnitude larger — to Cygames in games and AbemaTV in video. As the turning point where an ad-sales company changed its character into an operating company, this choice sits at the knot of the company’s lineage.
The meaning of this decision goes beyond immediate server reinforcement. For an ad agency — a sales-led company — to hire engineers itself and entrust the design and operation of its infrastructure to a technology chief was a decision to shift the centre of the company’s capabilities from sales to technology. As symbolised by a design that used MySQL and Oracle for different purposes, bringing things in-house was not merely securing manpower but creating a state in which it could optimise its own services with its own technology. Having decided to own media, it had to own the technology to support it too — the 2004 Ameba choice inevitably summoned this in-housing.
And the technological strength grown in this period did not stop at Ameba alone. An organisation able to design and run large-scale services in-house later became the parent body sending people and know-how into fields where technology decides competitiveness — the Cygames games business, and the ad-tech unit handling ad-delivery technology. The starting point at which CyberAgent, begun as a sales company, came to be recognised as a place for engineers to work also lies in this 2006 turn. Entry into media prompted the in-housing of technology, and that technology in turn bred the next business — the first link of this chain was tied here.
The feature of this turning point is that the trigger for reform was not a deterioration in results but a shareholder’s exercise of voting rights. Holding no stable shareholders, a base in which foreign corporations and institutional investors held much of the stock reflected the capital market’s assessment straightforwardly in calm times, but on the single point of board independence it also became a force that said no to the founder. The fact that, excluding Fujita’s own holding, a majority turned to opposition showed in numbers that the founder’s magnetism alone could not complete the governance of a listed company.
The company’s response was swift, and paradoxical. It voluntarily shrank the founder-led board that had been the source of that magnetism, and brought the outside eye close to a majority. That the next year’s approval returned to 95.65% confirms that what the market wanted was not Fujita’s exit but transparency of governance. And because a committee became the vessel for a succession plan, this affair did not end as mere firefighting but became the institutional starting point of the succession plan announced in 2023. Turning the external pressure of a shareholder revolt into a lever to push governance forward — the implication of this decision lies in that path.
The heart of this decision is that it did not measure additional investment in a loss-making business by “when it would be recouped.” Fujita named no deadline for turning profitable and instead built a structure in which two profit pillars — games and advertising — kept absorbing the media losses. On the strength of that stamina, he piled its largest-ever investment onto the World Cup, a one-off piece of major content, buying awareness and a user base rather than subscription revenue. Beginning with the financial capacity gained in the 2000 listing and continuing through Ameba and AbemaTV, the company’s management style — a profitable business supporting a loss-making one over the long term — is reproduced here again, an order of magnitude larger.
That said, the free live-stream rewriting the viewing records and the business turning profitable are not immediately connected. In the World Cup year the losses actually widened, and a full-year profit took about three more years from there. Even so, one cannot deny that the awareness and the lift in weekly active users won by making it free hastened the later swing to profit. In place of promising a time of recoupment, it concentrates investment all at once at the decisive moment — whether that judgement bears fruit is premised on the continuation of the high earnings of the core businesses that support it. The World Cup bet can be seen as a move played while that premise still held.
The heart of this decision is that the founder set the timing of his own exit not when cornered but in the midst of strong performance. While many founders cannot let go of their magnetism and delay succession, Fujita, seeing an internal document that imagined his future self, chose to move it up: he cultivated 16 candidates under the condition of being in their forties or younger over several years, and put the tacit knowledge of management into words as a handover document. Given the chain in which the 2019 shareholder opposition produced governance reform, and the committee that reform created became the vessel for the succession plan, this declaration was not sudden but a move placed on top of institution and preparation.
That said, it is early to conclude that the escape from founder-dependence has been achieved. The handover, initially set for 2026, was moved up to 2025, and Fujita continues for the time being with CEO-level involvement as chairman, accompanying the new structure under a two-representative system. Succession will truly be complete only when the company keeps growing after Fujita has left. Preparing to step back while riding high is itself one answer to founder-run management, which tends to postpone succession — but whether that answer was right will be proven by the second-generation structure from here on.
Each heading links to the full Japanese analysis — background, decision and outcome, with sources.
This is a condensed English edition. The full, source-by-source history — with the detailed narrative, financial tables, shareholders and executives — is maintained in Japanese: 日本語版（詳細）— CyberAgent full history in Japanese →