Original Author: Zhao Xuan

According to reports from The Wall Street Journal and other media, DeepSeek has recently completed its first round of large financing, with a financing scale of approximately $7.4 billion and a post-investment valuation exceeding $50 billion. Related reports also mention that Liang Wenfeng invested about $3 billion himself, and most external investors did not directly invest in DeepSeek's operating company, but entered through a limited partnership managed by the founder, which is subject to constraints such as a five-year lock-up period.

It should be noted that the above information currently mostly comes from media reports; DeepSeek has not yet publicly disclosed complete transaction documents. Therefore, this article does not comment on the authenticity of the transaction, nor does it attempt to restore DeepSeek's complete governance structure, but rather discusses a more general issue based on this media report: How can AI founders retain key decision-making power when introducing massive capital, provide reasonable protection to investors, and avoid turning the company's operational risks into their personal risks?

For the founding team, the truly dangerous question during financing is often not "How much am I diluted?" but rather: After the next round of funding, who can decide whether the company sells or not, whether to change the technology roadmap, whether to replace the CEO, whether core assets can be licensed out, and who will bear the buyback responsibility if the project fails, etc.

Why AI companies value control more than ordinary startups

AI companies typically require continuous substantial funding. Model training, inference computing power, core talent, data acquisition, and product commercialization can lead to long-term expenses. Relying solely on the founding team's own funds often makes it difficult to complete the entire process from technological research and development to large-scale application.

At the same time, the technology roadmap and business cycle of AI companies have significant uncertainty. Investors usually care more about revenue, valuation, and exit periods, while the founding team may be more concerned with model capabilities, open-source strategies, and long-term research and development. The two sides are not inherently in conflict, but at crucial nodes, their judgments may differ significantly.

The transaction windows in the AI industry may also be shorter than those in traditional industries. Once a product reaches a certain stage, it may need to be sold promptly to large technology companies for mergers or deep partnerships; it may also need to stop the original direction and quickly pivot to new models, scenarios, or business models.

These major decisions often cannot be discussed indefinitely. If the governance structure is too dispersed, every decision such as transformation may require repeated negotiations among multiple parties, causing the company to miss the optimal window.

Therefore, having people who genuinely understand the technology, products, and teams retain decision-making power over key matters helps the company make efficient decisions at critical moments. However, efficient decision-making does not mean that the founder can act without constraints. The value of control is to enable the company to make clear decisions, rather than allowing the founder to overlook the interests of the company, investors, and other shareholders.

Problems solved by limited partnerships

If investors directly invest in a startup, they usually become direct shareholders and may demand board seats, etc.

Limited partnerships offer an alternative arrangement: investors first invest funds into a holding platform, which then invests in the operating company. At the level of shareholders in the operating company, originally dispersed investors may be concentrated into a single holding entity.

Investors primarily gain economic benefits, informed supervision, and specific protections within the limited partnership (where informed rights and corresponding protections need to balance the interests of all parties); how limited partnerships exercise shareholder rights at the operating company level is defined by the partnership agreement.

If the transaction structure of DeepSeek disclosed by the media is accurate, this arrangement could reduce the degree to which external investors directly intervene in daily operations and technology routes, while enabling the founder to unify and coordinate decision-making within the holding platform.

However, limited partnerships are not a magic solution for control.

Controlling a holding platform only indicates that the founder may influence how one shareholder exercises rights, and does not necessarily mean that they have control over the entire company. True company control also depends on who decides key matters, who appoints and dismisses the core team, and whether these arrangements can remain in place through subsequent financing.

Control is not only related to equity

The founding team can determine whether they truly possess control by considering three questions.

The first question is, who ultimately decides matters that change the company's fate.

For example, whether the company continues to raise funds, sells or merges, transfers core models, codes, data, and intellectual property, changes the technology roadmap, or ceases existing business. Even if the founding team still holds a significant proportion of shares, if these matters can be unilaterally vetoed by investors, their actual decision-making space may still be significantly limited.

The second question is, who decides the core team.

Who can appoint or change the CEO, CTO, and core management personnel; who can decide the R&D budget and commercialization pace are often more impactful on the company's actual direction than the equity percentage itself.

The third question is whether the current control arrangements can be sustained.

Just because the founding team has decision-making authority today does not mean they will still have it after the next round of financing. The veto rights of new investors, changes in board seats, ongoing dilution of shares, and the mechanism for replacing executing partners within the holding platform (i.e., those with decision-making power) can all change the control situation.

Therefore, the founding team needs to design not an isolated holding platform but a complete decision-making chain: how funds enter through which entities, how those entities exercise shareholder rights, who decides significant matters, who appoints and dismisses the core team, and whether these arrangements will still be effective after the next round of financing.

Which decision-making powers should not be easily transferred?

The founding team does not need to have absolute decision-making authority over all matters. Investors' understanding of the company's financial situation, supervision of the use of investment funds, and restrictions on unreasonable related party transactions typically constitute reasonable protections.

What really needs to be cautiously transferred is decision-making power that could change the company's fate.

First is the disposal of core technology and intellectual property. Models, codes, training results, data resources, and core patents are usually the most important assets of AI companies. For the sale, exclusive licensing, or transfer of these assets, the founding team should at least retain joint decision-making rights or necessary veto rights.

Second is the company’s sale, merger, and major transformation.

Once AI products reach a certain stage, selling to large companies does not necessarily mean entrepreneurial failure; it may also be a reasonable path for the product, team, and technology to continue developing. However, when to sell, whether to sell equity or technology, and whether the founding team continues to retain involvement must be decided by someone who truly understands the company's long-term value.

Third is the appointment and dismissal of the core management team. If investors can easily replace the founder, CEO, or core technical leaders under regular circumstances, the founder's control may be merely formal.

Finally is subsequent financing and significant dilution. New financing rounds not only change ownership ratios but may also reallocate board seats, veto rights, and management appointment and dismissal powers. The founding team cannot just calculate the ownership ratio after this round of financing but should simulate the governance status after the next round or even the round after that.

The founding team should not broadly pursue "I decide everything." A more realistic approach is to first list three to five core decision-making powers that cannot be easily lost, and then provide investors with reasonable protections on financial oversight, information disclosure, and risk control.

Retaining control does not mean excluding investor protection

Investors willing to invest without directly controlling daily operations typically demand protection on other levels, such as understanding operational and financial situations, supervising the use of investment funds, restricting conflicts of interest, obtaining remedies in cases of serious breaches, and having exit options upon the attainment of reasonable deadlines or conditions.

The founder should distinguish between two types of arrangements.

One type is control over product direction, technology roadmap, and daily operations; the other type is rights to prevent asset transfer, conflicts of interest, and serious breaches.

The former can be retained more by the founding team, while the latter should leave reasonable space for investors. A mature financing structure does not mean that investors can only provide funds and watch from the sidelines but reduces their direct intervention in operational direction while ensuring their basic rights to be informed, supervise, exit, and seek remedies.

In several bet-related disputes I have dealt with, it is evident that in many cases the founding team did not provide investors with necessary rights to be informed and supervise, nor did they reserve reasonable exit mechanisms, while both parties signed stricter buyback, guarantees, pledges, and liability terms in search of compensation. On the surface, operational control was maintained, but in reality, it may have transformed the company’s operational risks into the founder's personal debt risks.

The more concentrated the control, the more the founding team needs to emphasize the company’s independence. Core intellectual property registered under personal or affiliated company names, personnel, data and R&D results mixed between multiple entities, and transferring funds and business opportunities through related-party transactions may turn the control issue into governance and personal liability issues.

Before signing the Term Sheet, founders should at least consider five questions

Valuation and dilution ratios are important issues, but financing negotiations should not be limited to these. Before signing the Term Sheet, the founding team should at least clarify the following five categories of questions.

The first is control

Which matters must be agreed upon by the founding team, which matters can be decided by a majority of the board, and which matters can be executed by the management team. It is especially important to specifically list the sale of the company, major financing, disposal of core assets, changes in technology roadmap, and appointment and dismissal of core management.

The second is investor protection

Which informed rights, supervisory rights, and exit rights can be granted to investors, and which veto rights may substantially affect daily operations. It is necessary to consider not just the names of the terms but the triggering conditions, applicable scope, and actual effects after layering different rights.

The third is buyback terms

Should buyback terms be accepted? What events may trigger the company's, founders', or actual controllers' buyback responsibilities; will failures to go public, performance shortfalls, subsequent financing failures, compliance issues, and founder departures all lead to buybacks; does the buyback price include fixed yield, compound interest, and high penalties.

The fourth is the holding platform

Who serves as the executing partner, who can replace that entity, and what supervisory, exit, and remedy rights do limited partners have. The holding platform must not only facilitate entry during financing but also consider whether it can be smoothly adjusted during subsequent financing, mergers, public offerings, and exits.

The fifth is dispute resolution

This is a very core agreement in practice that is often overlooked in the initial stages. Do investment agreements, shareholder agreements, buyback agreements, guarantee contracts, and holding platform agreements adopt mutually connected dispute resolution arrangements; when involving multiple entities, is there a fragmentation of procedures between partial litigation and partial arbitration.

The significance of these five questions is to translate the abstract concepts of "founder control" and "investor protection" into transaction conditions that can be discussed and modified item by item.

The more dangerous financing terms may not be equity ratios

In dozens of equity investment, buyback, and commercial arbitration projects I have participated in and handled, many founding teams focus primarily on valuation, received amounts, and dilution ratios during financing, while underestimating the long-term risks of buyback, guarantees, and dispute resolution clauses.

Buyback terms especially need to be treated with caution.

The founding team should confirm who exactly bears the buyback obligation—whether it is the company, the founding team, or the actual controller—and whether they bear joint liability; if the triggering conditions for the buyback are too broad; if the buyback price includes fixed yield, compound interest, and high penalties; and whether the founder has realistic abilities to fulfill this obligation.

Once the founder personally bears the buyback obligation, the risk of company project failure may directly convert into personal property risk. When the company bears the buyback obligation, it is also necessary not only to look at contractual stipulations but also to consider the company's capital maintenance rules and specific performance conditions.

Dispute resolution clauses must also not be handled hastily.

Disputes regarding equity investment, control, and buybacks often involve commercial secrets, multiple transaction documents, and several related entities. Commercial arbitration, characterized by its non-public hearings and relatively flexible procedures, is usually more suitable for handling complex commercial disputes.

However, choosing arbitration requires caution. The arbitration institution, location, applicable rules, arbitration language, and scope of the dispute should all be evaluated thoroughly before making a choice; otherwise, the founding team or investors may face difficulties and costs in enforcing rights that they did not fully understand when signing the agreement. In some chain investment disputes I participated in, the enforcement costs during arbitration exceeded the acceptable range for the rights enforcer, thus allowing the breaching party to significantly evade their liability.

Conclusion

For AI founders, financing documents are not merely a set of legal text formats, but rather an operating system that determines decision-making authority, risk-bearing, and exit paths for the company in the years to come.

A truly mature financing structure does not mean that founders are forever unrestrained, nor does it mean that investors can only watch; rather, it makes clear upfront: which matters will be decided by whom, which risks will be borne by whom, under which circumstances exits will be triggered, and which mechanisms will resolve disputes.

Before capital enters the company, control, buyback responsibilities, holding platforms, and dispute resolution mechanisms should all be concurrently designed. Otherwise, the day financing is completed may also be the day when future disputes over control and personal liability risks begin to accumulate.

免责声明：本文章仅代表作者个人观点，不代表本平台的立场和观点。本文章仅供信息分享，不构成对任何人的任何投资建议。用户与作者之间的任何争议，与本平台无关。如网页中刊载的文章或图片涉及侵权，请提供相关的权利证明和身份证明发送邮件到support@aicoin.com，本平台相关工作人员将会进行核查。