Разработка ПО
Менеджмент
Для компаний
Для стартапов
Архитектура ПО
Nvidia is losing ground in China, while Huawei is selling processors faster than it can produce them. Who will make money from this market reshuffle, and which niches are still free?
Оглавление
What happened
U.S. sanctions against Nvidia created a vacuum in the Chinese market for artificial intelligence accelerators. Huawei instantly filled this gap: according to the Financial Times, China's largest technology companies, Alibaba, ByteDance, and Tencent, are purchasing Ascend 910C chips en masse. Order volumes amount to tens of thousands of units. Huawei is expanding production capacity, but demand still outpaces supply. This is not just the replacement of one supplier by another; a fundamental shift is taking place in the architecture of the AI infrastructure of an entire region.
How this is useful for business
A market reshuffle always creates opportunities for those who find themselves between the old and new leader. Chinese companies forced to switch to Huawei face tasks: optimizing software for the new architecture, training staff, and integrating with existing systems. An entire layer of services is emerging that no one has yet occupied internationally. In addition, the chip shortage means companies will look for alternative ways to scale AI computing, creating space for cloud solutions, edge computing, and hybrid architectures.
How to make money from this
The first path is to become an integrator of Huawei-based solutions for Western companies that work with Chinese partners or enter Asian markets. The second is to develop software tools that optimize model operation for the Ascend architecture: compilers, migration frameworks, and monitoring systems. The third path is to create educational products: courses on working with Huawei AI Stack, certifications, and corporate training. The fourth option is infrastructure-focused: building edge computing nodes in regions with low coverage. The fifth is strategic positioning consulting: which companies benefit from switching to Huawei, and which should look for workarounds through other suppliers.
Business ideas
1. Migration service for AI projects. You offer a full cycle of transferring models from Nvidia CUDA to Huawei CANN: code audit, architecture adaptation, performance testing. Monetization is a fixed project rate of $15,000–50,000 depending on complexity, plus subscription support at $2,000/month.
2. Marketplace of optimized models. You assemble a library of open-source models adapted for Ascend, with performance verification. Revenue comes from premium access at $9–49/month and corporate licenses at $500+/month.
3. Edge computing for retail and logistics. You deploy a distributed network of edge servers with Huawei chips for on-site data processing: cameras, sensors, smart warehouses. Model: equipment rental at $300–800/node/month + maintenance service.
4. Educational platform on Huawei AI. Courses for developers, managers, and engineers. An advanced certification program costs $800–2,500 per course, with corporate packages starting at $10,000.
5. Consulting on the AI equipment supply chain. You help Asian manufacturers enter new markets and Western companies find reliable suppliers. Commission of 5–10% of the deal or a retainer of $5,000–15,000/month.
Risks and limitations
The main risk is political volatility. U.S. sanctions may expand to companies that help Huawei or work with Chinese clients. Technological risk: Huawei is actively developing its ecosystem, and competing with it directly at the hardware level is impossible. Legal risk: the dual-use nature of technologies creates gray areas in export regulation. Geographic risk: dependence on a single region makes a business vulnerable to logistics and currency fluctuations. The reality is this: China's AI chip market is not a separate ecosystem but part of a global system, and any changes in trade policy will hit all participants.
7-day action plan
Day 1–2: Study the Huawei CANN and Ascend Development Kit documentation. See which tools already exist and what is missing. Day 3: Make a list of 20 companies in your sector that work with Chinese partners or depend on Nvidia. Day 4: Conduct 5 calls with potential clients: identify migration pain points and willingness to pay for a solution. Day 5: Define a minimum viable product: it could be an audit report, a migration template, or a transition cost calculator. Day 6: Launch a landing page with an offer and collect a waitlist. Day 7: Analyze feedback and prioritize features for the MVP. Make a decision: which of the five monetization paths to choose and which resources to reallocate.
Original news: Financial Times Companies · See other news in the news section.
