Summary of Key Points
Liang Rubo, CEO of ByteDance, sent a company-wide letter after a four-year hiatus, updating the company's mission, management philosophy, and leadership principles, signaling ByteDance's full commitment to "All in AI." On the surface, this is an organizational mobilization for the AI era, but in reality, it aims to address the common issues of large companies (such as siloed departments, slow decision-making, and managers being detached from frontline operations) through the narrative of AI. However, there are underlying concerns with ByteDance's bet on AI: while its AI business data looks impressive, it may be inflated; its commercialization path faces limitations, and the allocation of resources to mature businesses could lead to a shortfall in revenue. Additionally, focusing on a single technological approach carries significant risks.
The True Intent of the Company-Wide Letter: A Deep Adjustment to Address Large-Corporate Problems with AI
The core of this letter is not just to shout slogans about AI but to tackle the specific challenges faced by ByteDance's workforce of 100,000 employees.
- Root of the Problems: Liang Rubo has previously acknowledged that ByteDance suffers from typical large-company ailments, such as departments working in isolation and slow response to AI transformations.
- Targeted New Leadership Principles:
- The addition of "aiming for high goals" and "daring to set ambitious objectives" reflects the company's shift away from past practices of reckless expansion and trying to do everything without specialization, aligning with ByteDance's recent efforts to narrow its business focus.
- "Maintaining a sense of crisis and going to the frontline" is emphasized as a way to reduce internal inefficiencies, such as unnecessary approvals and meetings.
- Potential Side Effects: These principles could lead to managers setting overly high expectations for their teams or requiring more reports from frontline staff in order to demonstrate their commitment, potentially increasing their workload.
The Truth About AI Business Data: Impressive but Inflated
The achievements of ByteDance's AI division, Volcano Engine, seem impressive, but upon closer inspection, they are somewhat inflated:
- Inflated Data: A 49.5% market share is based on a narrow metric (only model API calls, excluding traditional cloud services), with Alibaba Cloud leading in the full ecosystem.
- About half of the 180 trillion Token calls may be internal usage (for AI features within ByteDance's own products like TikTok and Jutoutiao), creating a self-supporting cycle.
- Revenue is only 20 billion yuan, half that of Tencent Cloud and one-fifth of Alibaba Cloud's—high call volumes do not necessarily equate to high profits, as it may be due to low pricing strategies.
- Major Weaknesses:
- The price war is unsustainable: ByteDance initially offered lower prices, but competitors like Alibaba and Tencent quickly responded (with price cuts on certain models and free entry-level models), ending the period of market dominance through cheap tactics.
- Weak infrastructure: AI requires substantial hardware resources, and ByteDance lags behind Alibaba and Tencent in this area. Corporate clients seek integrated services from compute power to models, which ByteDance can only partially provide with its MaaS model.
The Risks of "All in AI"
Betting all resources on AI means risking everything for the company:
- Shrinking Core Business: ByteDance's past revenue sources (advertising and e-commerce, especially through TikTok's shopping features) are now being diverted to AI. These mature businesses may lack investment, leading to slower growth.
- High Strategic Risk: The AI industry is still in its early stages, and the future direction of technology (whether it will be based on foundational models or hardware/specific use cases) is uncertain. By focusing solely on foundational models, ByteDance could become vulnerable if market trends change.
Commercialization Challenges
Both of ByteDance's revenue streams for AI face significant limitations:
- C-Side Subscriptions (DouBao Professional Edition): The free version meets most users' needs, and the paid additional features are not attractive to ordinary consumers. The low proportion of paying users means that subscriptions cannot sustain significant revenue.
- B-Side MaaS: Corporate clients require comprehensive services (compute power, customized models, maintenance), which ByteDance does not offer as well as Alibaba and Tencent, limiting its ability to attract and retain large customers.
Conclusion
ByteDance's AI transformation is a bold move driven by anxiety. While using AI to address internal issues is a positive step, the risks are clear: inflated data, difficulties in commercialization, shrinking core businesses, and high strategic stakes. Whether ByteDance can succeed with this all-in strategy depends on its ability to resolve these underlying challenges in the future.