Huawei’s Guo Ping on the New Strategy: No Big Models, but Becoming the “NVIDIA” of the AI World?
Hello everyone, I’m your financial analyst.
Recently, an important document circulated within Huawei – a summary of a discussion between Guo Ping, the chairman of the supervisory board, and new employees. This document contains a wealth of information that not only reveals Huawei’s strategic ambitions in the AI era but also directly addresses various speculations about Huawei’s chips, business scope, and approach to talent management.
Many non-experts might find terms like “Shengteng,” “Kunpeng,” and “Tao’s Law” confusing. Don’t worry; today, we’ll break down these technical jargon into plain language to understand what Huawei really wants to achieve and what it means for us ordinary people.
Summary of the Key Points
In short, Guo Ping’s speech has defined Huawei’s new role in the AI era: Huawei does not aim to be the smartest “brain” (big model) but rather the strongest “body” (computing infrastructure). Huawei’s goal is to become the “NVIDIA” of the AI industry. This means that no matter which company’s big model is the most advanced, Huawei ensures it runs fastest and most reliably on its chips and servers. At the same time, Huawei has made it clear that it will not blindly expand its business scope but will continue to focus on “connection” and “computation.” Despite chip limitations, Huawei plans to compensate for these shortcomings through architectural innovation and a strategy of “using time to gain space.” Finally, Guo Ping reminded employees that in the AI era, rote memorization is no longer enough; the ability to ask questions, solve problems, and continuously learn is what matters most.
---
In-Depth Analysis: Understanding Huawei’s AI Ambitions from Five Dimensions
1. Strategic Positioning: Not Being a “Chef,” but Creating a “Michelin-Level Kitchen”
[Plain Language Explanation]
People used to ask whether Huawei would develop its own big models like ChatGPT. This time, Guo Ping was clear: Huawei’s core goal is not to create the most powerful big model, but to ensure that all big models can run efficiently on its platforms.
Think of it like running a restaurant:
- Old approach: Develop the most complex recipes (create big models) and become the best chef.
- New approach: Build the best kitchen in the world (with technologies like Shengteng, Kunpeng, and super-node clusters). Whether you’re a Michelin-starred chef or a street vendor, as long as you cook in our kitchen, we guarantee precise cooking, fast service, and low costs.
Guo Ping mentioned that Huawei wants to be the “NVIDIA” in the ICT (Information and Communication Technology) and computing fields. Why is NVIDIA strong? Not because it develops the best AI applications, but because 90% of AI training is done on NVIDIA’s GPUs. Huawei’s strategy is to provide the best computing power, regardless of which big model is used. This is a pragmatic and commercially wise move that avoids direct competition with big model manufacturers and allows Huawei to dominate the infrastructure market.
2. Business Scope: Focusing on What Matters
[Plain Language Explanation]
Many companies try to diversify their businesses (phones, cars, finance, etc.). But Guo Ping stated clearly that Huawei has no plans for expansion. This means Huawei will focus on its core areas: “connection” (5G, optical networks) and “computation” (chips, servers, cloud computing).
Why? Guo Ping explained: “What really matters is our own capabilities, not the external environment.” It’s like a martial artist; instead of learning a hundred useless techniques, it’s better to master one skill to the highest level. In the fast-paced AI world, diversifying can lead to falling behind. Huawei’s strategy is to strengthen these two pillars, and other businesses (such as smart cars and devices) will naturally benefit from this.
3. Chip Breakthrough: Finding a Way Around Limitations
[Plain Language Explanation]
This is a major concern for Huawei, as it faces chip manufacturing restrictions due to US sanctions. Guo Ping mentioned “Tao’s Law,” which suggests a different approach:
- Traditional method: Trying to make the chips smaller (like building narrower roads with more lanes). This requires advanced processes (3nm, 2nm), which Huawei lacks.
- Huawei’s method: Instead of making the chips smaller, Huawei focuses on optimizing software, architecture, and systems to make data flow more efficiently within the chips, reducing waiting times.
Huawei plans to integrate design and manufacturing to compensate for process shortcomings. In other words, even if its chips are not as advanced as others’, its chip architecture is more sophisticated, ensuring better performance and reliability.
Regarding the Shengteng 950 series, Guo Ping emphasized that it’s not about simply increasing parameters (number of cores, frequency) but about “solving real problems.” Huawei’s chips may not outperform others in individual metrics, but they will have an overall advantage in cluster efficiency, stability, and cost-effectiveness through technologies like super nodes. This is about leveraging strengths and avoiding weaknesses.
4. AI Implementation: From “Showoff” to “Practical Use,” with Operators as the First Test Bed
[Plain Language Explanation]
AI is popular, but many applications are still in the “chatting” phase. Guo Ping pointed out that AI needs to be integrated into traditional industries, such as operator networks. Imagine a 5G network with many devices and complex connections; manual maintenance is inefficient. In the future, AI will automatically detect, repair, and optimize network issues.
This requires “practical AI talent” that understands networks, business processes, and can use AI tools to solve real problems. Huawei also distinguishes between different AI applications:
- ICT/computing: Helping customers build models (infrastructure).
- Endpoints/smart driving: Developing custom models for specific scenarios.
- Huawei Cloud: Developing its own models for cloud services, as models are crucial for cloud competitiveness.
This shows Huawei’s clear understanding of AI applications: it knows where to develop in-house, where to collaborate, and where to outsource.
5. Talent Management: In the AI Era, Asking Questions Is More Important Than Knowing Answers
**[Plain Language Explanation]
Guo Ping discussed that AI could be the last major technological revolution, which will change the workplace. In the past, we judged people by how much they knew and how fast they could calculate. Now, AI knows more and can calculate faster. The new skills for success are:
1. Clearly defining problems: Ask the right questions.
2. Continuously questioning: Evaluate and correct incorrect answers.
3. Solving real problems: Focus on achieving results.
4. Continuous learning: Adapt to constant changes.
For us, this means that AI won’t replace us, but those who can use it effectively will. The future’s core competencies are not just knowledge but the ability to apply it and think critically.
---
Conclusion
In summary, Guo Ping’s speech reflects Huawei’s “calm and pragmatic” approach to the AI era. Instead of blindly chasing big models, Huawei focuses on its strengths in connection and computation, innovates in architecture, and avoids resource dispersion. This strategy is about endurance and systematic capabilities, not speed or single-point breakthroughs.
For us, the biggest change brought by this technological revolution might not be smarter phones but a shift in our way of thinking: from finding answers to defining problems. This is the real key to success in the AI era.