Chen Jingjing on China’s AI Stack: From Cheap Energy to Chip Self-Sufficiency

TigerClub
16:52
Reward Tiger-CoinsReward 500 Tiger-coins

Speaker: Chen Jingjing of ChinaAMC maps China’s AI industry as a five-layer cake of energy, chips, infrastructure, models and applications, then zooms in on the chip supply chain at its centre.

[About the Speaker] Ms Chen looks after investors and products in the APAC region at ChinaAMC. She has over 8 years of industry experience and joined ChinaAMC in 2024, following roles at E Fund Management and HSBC in the UK, where she specialised in Global Banking, Leveraged Finance and Sustainable Finance. She holds a Master’s in Management from the University of Cambridge and dual degrees in Journalism and Economics & Management from Tsinghua University.

[The Hook] Jingjing opened with a show of hands. Everyone knew NVIDIA. Everyone knew SK Hynix and the Korean and Japanese AI names. Chinese AI companies were the least familiar of the three, even though, she said, China has leading players across the AI chain. Her five-layer cake was how she set out to close that gap.

[Who Is ChinaAMC?]

Jingjing introduced her firm briefly. ChinaAMC is headquartered in Beijing and manages about US$463 billion. It has an investment team of more than 300 people who talk to policymakers and company owners on the ground. It issued China’s first ETF about 20 years ago, and today it is China’s largest ETF issuer with about 20% market share. China is already the largest ETF market in Asia.

[The Five-Layer Cake]

Layer 1 — Energy: China leads. Energy powers all AI infrastructure, and China invested in it long before the AI race began. Jingjing’s figures: energy costs are about 30% lower than in the US, installed electricity capacity is about three times the US level, and network infrastructure costs are about 60% lower. The US, by contrast, already faces electricity constraints.

Layers 2 and 3 — Chips and infrastructure: the US leads, but China matters. The US has the most advanced chip designs and the hyperscalers. China is still an important part of the global AI supply chain, especially in optical modules, which connect GPUs inside data centres. Seven of the world’s top ten optical module makers are Chinese, and about 70% of their revenue comes from overseas, so they benefit from US AI spending too. The bigger story, she said, is self-sufficiency: NVIDIA’s share of China’s AI chip market was about 80% two or three years ago, is about 30% now, and she expects it to fall to single digits by 2027, in a market of more than US$50 billion in revenue.

Layer 4 — Models: cheaper and open source. Chinese models have not surpassed US models, she said. What got attention was that they deliver comparable performance at about one-third to one-fifth of the cost of leading US models, and they are open source.

Layer 5 — Applications: where she expects China to win. Cheaper energy, a more self-sufficient chip supply chain and cheaper models all lower the cost of AI, and a lower cost speeds up adoption. Two examples she gave: an electric vehicle is “a traditional car plus AI”, and industrial and humanoid robots are manufacturing plus AI.

[Chips, Layer by Layer]

When the US began restricting exports of advanced chips to China, she said, local companies first panicked, but it also gave them a window to grow. Where the chain stands now:

  • AI chip design — $NVIDIA(NVDA)$ is the global leader. Its Chinese counterparts include Cambricon, MetaX and Huawei, and she said they are closing the gap at a very fast speed.

  • Memory — $Samsung Electronics Co., Ltd.(SSNLF)$ and $SK hynix(SKHY)$ lead in DRAM, and Chinese memory makers are catching up quickly.

  • Foundry — TSMC leads. SMIC is China’s main player, and the speakers acknowledged it is still behind TSMC in chip technology, although it has improved. Mature-node chips are used more for industrial purposes, while advanced nodes serve AI training. Jingjing said Chinese makers can now produce chips at around 4 nanometres.

  • Equipment and EDA software — the hardest part. Lithography is a key constraint in China. Companies such as Naura and AMEC are trying to catch up, and she expects some of them to become world leaders one day.

[Case Study: Cambricon]

Jingjing’s first example is China’s closest equivalent to $NVIDIA(NVDA)$:

  • Revenue grew about 400% in 2025, partly because of a low base in 2024, and about 100% again in 2026.

  • It only turned a profit in 2025, after years of burning cash, and it is a key supplier to Chinese cloud and internet companies, including Alibaba.

  • It trades at about 130 times earnings. She expects that to come down to around 30 times by 2027 if growth continues. That is her estimate.

The lesson: a high P/E means high expectations, and the stock only makes sense if the growth shows up.

[Case Study: A Chinese Memory Maker]

Her second example is a Chinese memory-chip company that competes directly with Samsung and SK Hynix. It listed in June, and its shares jumped about 400% on the first trading day. Its global market share is about 4%, and it is trying to win share overseas as well. Two education points came out of this. The ETF took part in the IPO, so it held the stock from listing day. And a newly listed name only joins the index after a monitoring period of about six months.

[The Chip ETF]

The semiconductor ETF holds 30 names across the supply chain. The China-listed ETF behind it launched about six years ago, now holds about US$4 billion, and trades about US$200 million a day. Jingjing was frank about the behaviour: it can rise 5% one day and fall 4% the next. In her view, long-term buy-and-hold is not the right way to approach China markets. The volatility is the nature of the theme, and she suggested trading it. She also sees China semis and US tech as complementary exposures rather than substitutes.

[Key Takeaways]

  • AI is a stack, not one stock. The five layers have different leaders, and China is strongest in energy and applications.

  • Cost is the engine. Cheaper energy and models lower the cost of AI, which speeds up adoption.

  • Self-sufficiency drives the chip story. Some layers are further along than others, and lithography and EDA software are still bottlenecks.

  • High growth often comes with a high valuation. Cambricon’s 130 times P/E shows how much is already priced in.

  • Volatility is part of the theme. Size your position for it.

[Call to Action 🎯]

  1. Take one AI company you already follow and place it on the five-layer cake. Which layer is it in?

  2. List the parts of the chip supply chain (design, memory, foundry, equipment) and name one listed company in each.

  3. Drop a comment: which layer do you think will matter most in ten years?

Markets are always moving - and sometimes, the best move is knowing what works for you.

With Treasury yields, oil prices and rate expectations keeping markets on edge this week, investors are once again thinking carefully about where to position next. There’s no one-size-fits-all choice in investing — and the same goes for Tiger Merch. This month’s hot picks are in, featuring the Tiger Toiletry Bag, Universal Travel Adapter, Tiger Umbrella and more favourites chosen by fellow Tigers.

Explore the Monthly Hot Picks in Tiger Coin Mall, now 12% OFF for a limited time.

https://laohu8.com/J/redeemGift?goodID=100561&type=delivery

Disclaimer: Investing carries risk. This is not financial advice. The above content should not be regarded as an offer, recommendation, or solicitation on acquiring or disposing of any financial products, any associated discussions, comments, or posts by author or other users should not be considered as such either. It is solely for general information purpose only, which does not consider your own investment objectives, financial situations or needs. TTM assumes no responsibility or warranty for the accuracy and completeness of the information, investors should do their own research and may seek professional advice before investing.
Reward 500 Tiger-coinsDeadline to 11/29 16:52
The originator will select the best recovery and allocate Tiger coins before the reward ends
Reward-post

Comments

  • Jerry Lam
    17:21
    Jerry Lam
    🐯 **我的选择:十年后,AI产业链最重要的可能是第五层——应用层,但第二层芯片仍然是最关键的技术壁垒之一。**

    **① 芯片决定AI能力的上限,应用决定AI商业价值的规模。** NVDA、TSM等企业掌握关键算力技术,但最终决定AI投资能否持续的,是企业和消费者能否通过AI创造真实经济价值。

    **② 中国的优势可能在AI规模化应用。** 相对较低的能源和模型使用成本,加上庞大的制造业基础,有利于工业机器人、智能汽车和自动化系统落地。我尤其看好AI与实体经济结合的长期机会。

    **③ 国产半导体替代具有潜力,但不能忽视技术瓶颈。** 寒武纪、中芯国际及国产设备企业值得关注,不过先进制程、光刻和EDA仍存在挑战。国产替代带来增长空间,不代表所有公司都能实现高利润。

    **④ 高增长不等于高投资回报。** 如果一家企业已经达到130倍市盈率,未来盈利必须持续兑现才能支撑估值。相比追逐短期涨幅,我更关注收入增长、毛利率、研发投入和自由现金流。

    🚀 **我的策略:美国科技股关注先进算力与核心技术,中国科技股关注国产替代和AI应用落地,两者互补配置,并控制主题仓位。**

    **结论:未来十年,芯片是AI发展的基础,应用才是检验AI经济价值的最终战场。**

    **我更看好那些不仅能制造先进技术,还能真正降低成本、提高生产效率,并持续创造利润的企业!📈**

Leave a comment
1
2