Is it still worth investing in AI stocks in 2025? A comprehensive analysis of AI concept stocks you must read before investing

The explosive popularity of generative AI has passed the initial frenzy. From the emergence of ChatGPT to now, the performance of AI concept stocks has long been a tale of two extremes—some have doubled, while others have fallen from high levels. So, entering 2025, do AI concept stocks still have investment value? How should one choose? These questions are worth a good discussion.

Why AI Concept Stocks Are Still Worth Paying Attention To

First, the conclusion: AI concept stocks still have opportunities in the short term, but precise selection is necessary.

According to the latest IDC forecast, global enterprise spending on AI-related technologies and solutions will reach $307 billion in 2025. Even more astonishing, by 2028, this figure will double to $632 billion, with a compound annual growth rate (CAGR) of 29%. This is not just a number; it reflects real enterprise investment demand.

The most critical point: The proportion of accelerated server spending will exceed 75% in 2028. This means that the hardware infrastructure supporting AI operation remains the biggest current dividend source. As a result, from chip manufacturers to server suppliers and cooling solution providers, the entire industry chain is benefiting from this wave of dividends.

How to Classify AI Concept Stocks? What Opportunities Do They Have?

The AI industry chain can be roughly divided into three layers: chip layer, application layer, and infrastructure layer.

Chips and accelerators are currently the most关注 track. NVIDIA dominates this field with its GPU, but AMD, Supermicro, and others are also eyeing this space. In Taiwan, TSMC, as the world’s largest wafer foundry, produces almost all high-end AI chips; MediaTek is actively布局 in edge AI and edge computing.

Servers and infrastructure are the second hot area. Quanta Computer’s Quanta Cloud Technology (QCT) has become one of the world’s largest AI server suppliers, with market share continuously expanding. Meanwhile, the surge in cooling demand has led to the rise of liquid cooling solution providers like DoubleHawk— as AI chips’ power consumption surpasses 1 kW, traditional cooling can no longer handle it.

Application layer companies include cloud platform giants like Microsoft and Google, as well as specialized AI chip design service providers like Weichips-KY. These companies have greater growth potential, but competition is fierce, requiring stronger commercial monetization capabilities.

Which Taiwan AI Concept Stocks Are Worth Watching in 2025

Quanta (2382): Transitioned from notebook OEM to AI server manufacturer. In 2024, revenue reached NT$1.3 trillion, with the proportion of AI servers continuously rising. In 2025, Q2 revenue surpassed NT$300 billion, up over 20% year-on-year. Foreign institutions set target prices between NT$350 and NT$370, with room for further growth.

Weichips-KY (3661): Specializes in customized chip design, serving cloud giants and AI leaders. In 2024, revenue exceeded NT$68.2 billion, up over 50%. By Q2 2025, quarterly revenue doubled to NT$20 billion, with gross margin continuing to improve. Foreign targets are NT$2200–NT$2400, with significant potential.

Delta Electronics (2308): A versatile player in power management and AI server cooling. In 2024, full-year revenue was NT$420 billion, with data center and AI-related income steadily increasing. In Q2 2025, revenue reached NT$110 billion, up over 15% year-on-year, benefiting from the global data center boom.

MediaTek (2454): A major fabless chip designer, actively developing in mobile AI and automotive AI. In 2024, revenue was NT$490 billion; in Q2 2025, NT$120 billion, up 20% year-on-year. Foreign targets are NT$1300–NT$1400.

TSMC (2330): The backbone of global chip manufacturing, producing nearly all advanced chips. As the behind-the-scenes contributor to AI chips, it remains the most stable long-term beneficiary.

Which US Stocks Are Worth Watching in AI Concept Stocks

NVIDIA (NVDA): The absolute leader in AI chips. In 2024, revenue was $60.9 billion, up over 120%. In Q2 2025, revenue reached $28 billion, with net profit increasing over 200%. The launch of the Blackwell architecture further consolidates its position in the AI market. Institutions are generally optimistic, raising target prices.

AMD (NASDAQ: AMD): A challenger to NVIDIA in the AI chip market. In 2024, revenue was $22.9 billion, with data center business up 27%. In Q2 2025, revenue grew 18%, with the MI300 series adopted by major cloud providers.

Broadcom (AVGO): Supplier of customized chips and network connectivity solutions. In FY2024, revenue was $31.9 billion, with AI-related products accounting for 25%. In 2025, demand from cloud providers for its AI chips and switches continues to rise.

Microsoft (MSFT): A leader in enterprise AI monetization. In FY2024, revenue was $211.2 billion, with Azure cloud services growing 28%. Its exclusive partnership with OpenAI and enterprise deployment of Copilot make Microsoft the most certain beneficiary of AI commercialization. In Q1 2025, intelligent cloud revenue first exceeded $30 billion.

Google (GOOGL): Although initially passive in generative AI, leveraging its technology and user base, it is accelerating to catch up. Overall, a long-term promising target.

Three Ways to Invest in AI Concept Stocks

Method 1: Directly buy stocks
Advantages: easy trading, low transaction costs, but risks are concentrated, requiring stock selection skills. Suitable for investors with some understanding of the AI industry.

Method 2: Buy equity funds
Fund managers select a portfolio of stocks for you, diversifying risk, but management fees are higher, and transaction costs increase. For example, First Gold Global AI Robot and Automation Industry Fund.

Method 3: Buy AI-themed ETFs
Lowest trading costs, passive index tracking, avoiding stock picking hassles. In Taiwan, there are products like Taishin Global AI ETF (00851), Yuan Da Global AI ETF (00762), among others.

Data shows that by the end of Q1 2025, the total assets of global AI and big data funds exceeded $30 billion, indicating increasing institutional and individual investment in this field.

Beware of This Investment Trap

While the outlook for AI concept stocks seems bright, history offers lessons. During the last internet bubble, Cisco Systems (CSCO) surged to $82, then fell over 90%, and 20 years later, it still hasn’t recovered to its high.

Similar patterns appeared in downstream internet companies. Yahoo was once a leader but was replaced and delisted by Google; Microsoft and Google’s stock prices also experienced long-term downturns after peaking in bullish markets.

Therefore, the investment logic for AI concept stocks is simple: initial benefits go to infrastructure and chip companies, but high growth is hard to sustain. In the medium term, if application companies can truly monetize AI technology, they will have long-term investment value. But even industry leaders are not immune to valuation corrections.

How to Rationally Invest in AI Concept Stocks in 2025

First, avoid chasing highs. Just look at Bridgewater’s moves— they are increasing holdings in NVIDIA, Google, Microsoft, but also managing risks, not blindly investing in everything.

Second, diversify. Use dollar-cost averaging to reduce costs, and combine stocks, funds, and ETFs to avoid risks from single companies.

Third, focus on earnings growth. Over time, market enthusiasm for AI will become more rational. The key is whether companies’ actual profit growth is still accelerating, not just hype.

Fourth, monitor policy changes. Countries regard AI as a strategic industry, likely increasing subsidies and infrastructure investments, which is positive. But tightening regulations on data privacy, algorithm bias, and other norms also pose risks.

Outlook for AI Concept Stocks from 2025 to 2030

In the short term, infrastructure providers like chip and accelerated server suppliers will remain the biggest beneficiaries. In the medium to long term, AI applications in healthcare, finance, manufacturing, autonomous driving, and other industries will gradually mature, becoming the true growth drivers.

Overall, from 2025 to 2030, the investment landscape for AI concept stocks will feature “long-term optimism with short-term volatility”. Macroeconomic factors (interest rate policies), regulatory environment, and market enthusiasm will influence stock price fluctuations. Therefore, the most prudent strategy is to prioritize hardware infrastructure and companies with tangible application deployment, adopt long-term allocation, and stagger entry points rather than chasing short-term gains.

For individual investors, diversifying through AI-themed ETFs is also a good choice. After all, in this rapidly changing field, even institutions find it hard to time the market perfectly; spreading risk is more important than betting on the right timing.