Rocket Internet, the leading player on Berlin’s tech scene, is planning the biggest initial public offering in the history of European Internet business. The data the company has had to release before the flotation, however, reveals it as a risky venture in which the IPO appears to be a last chance to preserve the underlying business model.

Rocket, founded in 2007 by the brothers Oliver, Marc and Alexander Samwer, is a startup factory. It takes successful US business models, such as those of eBay, Zappos, Groupon and Airbnb, and ports them to markets where the models’ originators are weak or absent. The company prides itself on taking startups from idea to public launch in just 100 days. The recent rollout of apartment cleaning service Helpling, a HomeJoy clone, in Germany took just 80 days.

These results are achieved by driving teams relentlessly and sometimes ruthlessly, as reflected in the company’s reviews on the employment site GlassDoor: “You either deliver (even bad code) or you’re fired"; “Slightly cold and harsh environment. Constant focus on ‘execution’ can be tiring at times’"; “be ready to deliver from day one." The churn rate is high.

The end result — or, rather, the core of Rocket’s current value, set at $8 billion at the midpoint of the IPO’s pricing range — is a constellation of what the company calls “proven winners." These are mainly fast-growing electronic retail companies launched in the last three years, leading to this somewhat paradoxical statement in the prospectus: “Our proven winners generated aggregated net losses of €442 million" ($568 million).

Rocket “winners" (Dafiti, Lamoda, Namshi, Jumia, Zalora, Jabong) sell clothing, shoes and accessories in Latin America, Russia, the Gulf states, Africa, Southeast Asia, India, Australia and New Zealand — places where the big US retailers are either absent or do not dominate the market. The Rocket-produced companies are losing money but growing revenues and customer numbers. That alone wouldn’t be unheard of in a tech IPO, but Rocket’s story has an added twist.

The “mother ship" itself owns stakes of between 21.4% and 49.5% in the 11 “winners." They are, therefore, not consolidated into Rocket’s income statements. The holding company’s revenues come from two sources: the sales of smaller e-tailers in which it owns controlling interest — a mere $60 million in the first half of 2014 — and something called “income from associated companies." 1 Those are not dividends, because the companies in which Rocket holds shares make no profit. The income accrues when they attract a fresh funding round, which dilutes Rocket’s stake but sometimes disproportionately raises the startups’ valuations. In the first half of 2014, that added about $17 million in nominal value to Rocket’s bottom line.

The Berlin incubator has been able to attract adventurous investors, such as the Swedish group Kinnevik, which has an 18% stake in Rocket. In May, the Swedes received “dividends in kind" from the German company, increasing their stakes in Rocket’s portfolio companies. They have also traded part of their Rocket stake for shares in the portfolio companies. As Kinnevik chief executive officer Lorenzo Grabau explained on a recent earnings call, “the models that have been built were very capital-intensive e-commerce businesses. The dilution we would have suffered by just investing at the Rocket level and not in the investee companies would have been such that we would have owned very little of that."

As a result of this investor interest in specific Rocket projects, rather than in the incubator itself, funding further growth the way the Samwer brothers have been doing until now would soon force Rocket to take its “winners" off the books altogether: Stakes of less than 20% are not considered significant for accounting purposes. Hence the IPO: some of the $1.9 billion the Samwers hope to raise with Rocket’s flotation will be used to increase their stakes in the portfolio companies.

E-commerce, as Grabau pointed out, requires big upfront investments. While the Samwers’ idea of serving the world outside the US and China is perfectly reasonable, the capital requirements could have forced them to relinquish control of the project and they want to rectify that. Investors will probably bite, despite the drawbacks: Recent tech IPOs show they are more interested in exciting stories than in revenues or safe, solid corporate governance structures.

1 Here’s how this is explained in the prospectus: “For associated companies accounted for using the equity method, if the effect of dilution on the Issuer’s interest is less than the overall increase in equity in the relevant company, the Group’s consolidated net income will show a gain in the line item “Income from associated companies" in the Group’s consolidated income statement, reflecting the increase in the Group’s share of the equity of the associated company. For example, in 2013 the entire volume of financing rounds in Bigfoot I, Bigfoot II and BigCommerce amounted to €220.5 million, €107.1 million and €91.8 million, respectively. The volume of these financing rounds multiplied by the Issuer’s stakes in Bigfoot I, Bigfoot II and BigCommerce corresponds to €51.5 million, €29.6 million and €37.1 million, respectively, being included in income/loss from associated companies in the Group consolidated financial statements, reflecting the relevant proportions attributable to the Group and dilution effects." Bloomberg