Who is benefiting from the AI boom? 2025 US and Taiwan Stock AI Concept Stock Deployment Guide

Gold Rush in the AI Wave

Since the explosive popularity of ChatGPT, AI is no longer just a future concept on paper; it is actively transforming the entire capital market. According to the latest IDC data, global enterprise spending on AI solutions and technologies is expected to reach $307 billion by 2025, surpassing $632 billion by 2028, with a compound annual growth rate of about 29%.

This is not just a numbers game. What does it represent? Every GPU working overtime, data centers expanding, and chip factories operating at full capacity. The surge in demand is creating a group of listed companies with significant profit growth.

The True Portrait of the AI Industry Chain

To understand why AI concept stocks are hot, we need to clarify the structure of the AI industry. The industry chain roughly consists of three layers:

Infrastructure Layer: Provides hardware support such as chips, servers, and cooling. This is essential for AI deployment, with the most direct and vigorous demand. According to IDC forecasts, by 2028, the expenditure on accelerated servers will account for over 75%, highlighting the importance of hardware suppliers.

Platform Layer: Cloud computing, AI operating systems, development tools, etc. This layer acts as a bridge between technology and applications.

Application Layer: AI applications across various industries. Growth here is relatively lagging but has the largest long-term potential.

From an investment perspective, the initial beneficiaries are inevitably infrastructure suppliers. This also explains why chip stocks, server stocks, and cooling stocks have soared in recent years.

The Four Main Players in US AI Concept Stocks

NVIDIA (NVDA): The Absolute Dominator of AI Chips

It’s hard to imagine a company holding such a position—from training large models to deploying inference, NVIDIA’s GPUs are almost the only choice. In 2024, revenue reached $60.9 billion, with an annual growth rate of over 120%. In Q2 2025, revenue hit a new high of $28 billion, with net profit increasing over 200% year-over-year.

This is not a false boom. Orders for Blackwell architecture GPUs (B200, GB200) are booked into next year, and cloud service providers are rushing to purchase. Analysts generally believe that as AI shifts from training to inference, this company’s growth cycle will continue, with multiple institutions setting target prices above $600.

Broadcom (AVGO): The Hidden Champion

If NVIDIA is the brain of AI, then Broadcom is the connector of AI neural networks. The company holds a monopoly-like position in custom ASIC chips, network switches, and optical communication chips.

In fiscal year 2024, revenue was $31.9 billion, with AI-related products rapidly increasing to 25%. In Q2 2025, revenue grew 19% year-over-year, benefiting from cloud providers accelerating the deployment of AI data centers. Products like Jericho3-AI chips and Tomahawk5 switches continue to see rising demand. Market expectations suggest that this company’s growth has just begun, with target prices often above $2,000.

AMD (Advanced Micro Devices): The Challenger’s Counterattack

NVIDIA has monopolized the market for too long, prompting AMD to adopt more aggressive strategies to break the deadlock. After launching the Instinct MI300 series accelerators, AMD successfully snatched some orders from NVIDIA. In 2024, data center revenue increased by 27% year-over-year, and in Q2 2025, revenue rose 18% year-over-year.

The key is that the MI350 series will be launched in the second half of the year, further eroding NVIDIA’s market share. Market analysis indicates that AMD’s AI chip market share still has huge room for growth, with target prices mostly above $200.

Microsoft (MSFT): The Pioneer in Enterprise AI Monetization

If the first three companies earn from hardware, Microsoft earns from applications. Its exclusive partnership with OpenAI gives Microsoft the largest access point for enterprise AI monetization.

In FY2024, revenue reached $211.2 billion, with Azure and other cloud services growing 28%, and AI services contributing over half of that growth. In FY2025, intelligent cloud revenue first surpassed $30 billion. Large-scale deployment of Copilot for Microsoft 365 and exponential growth of Azure OpenAI services make this company the most certain beneficiary of the “enterprise AI popularization” wave. Institutional target prices range from $550 to $600.

The Five Major Targets of Taiwan AI Concept Stocks

Quanta (2382): From OEM Manufacturer to AI Server New Star

Quanta was once the world’s largest notebook OEM manufacturer, now it is a key part of the AI server supply chain. Its subsidiary QCT specializes in servers and cloud solutions, serving clients including NVIDIA and international cloud providers.

In 2024, revenue reached NT$1.3 trillion, with the proportion of AI servers continuously increasing. In Q2 2025, revenue exceeded NT$300 billion, up over 20% year-over-year, setting a new historical high for the same period. Analysts are generally optimistic about its long-term growth, with foreign institutional target prices averaging NT$350–NT$370.

Unigroup-KY (3661): The Hidden Champion in ASIC Chip Design

Unigroup focuses on ASIC chip design services, with clients including major US cloud service giants and leading HPC and AI companies. In 2024, full-year revenue was NT$68.2 billion, up over 50%. In Q2 2025, single-quarter revenue surpassed NT$20 billion, doubling compared to the same period last year.

Behind the continuous improvement in gross margin and net profit margin is the entry of large AI clients into mass production, with new generation AI accelerators and data center orders arriving one after another. Foreign institutional target prices range from NT$2,200 to NT$2,400.

Delta Electronics (2308): From Power Management to AI Infrastructure

Delta Electronics is a global leader in power management, actively entering the AI server supply chain in recent years, mainly providing high-efficiency power supplies, cooling, and cabinet solutions. In 2024, full-year revenue reached NT$420 billion, with the proportion of data center and AI-related performance continuously rising.

In Q2 2025, revenue exceeded NT$110 billion, up over 15% year-over-year, benefiting from expanding demand for AI servers and data center infrastructure, with gross margin remaining high.

MediaTek (2454): Dual Engines of Mobile AI and Automotive AI

MediaTek is among the top ten fabless semiconductor design companies worldwide, actively advancing AI chip layout. Its Dimensity series mobile platforms have built-in enhanced AI computing units, and it collaborates with NVIDIA to develop automotive and edge AI solutions.

In 2024, full-year revenue reached NT$490 billion, and in Q2 2025, revenue was NT$120 billion, up about 20% year-over-year. The growth is mainly driven by increased market share in high-end mobile chips and rising demand for AI smart devices. Institutional target prices range from NT$1,300 to NT$1,400.

Sunway (3324): The Winner in Liquid Cooling for the AI Era

As AI server chips’ power consumption surpasses the kilowatt level, traditional air cooling has reached a bottleneck. Sunway leverages leading liquid cooling technology to secure a position in the global AI server supply chain. In 2024, revenue was NT$24.5 billion, up over 30% year-over-year.

In 2025, benefiting from major cloud providers accelerating the adoption of liquid cooling solutions, shipments of water-cooled AI server modules surged from Q2 onwards. With the launch of higher-power AI accelerators, the penetration rate of liquid cooling will rapidly increase, and foreign institutional target prices are mostly above NT$600.

Three Paths to Layout AI Concept Stocks

Direct Stock Purchase: Suitable for investors with in-depth understanding of specific companies. Risks are concentrated, but trading is convenient with no extra management fees. Leading stocks like NVIDIA, TSMC have ample liquidity and the lowest transaction costs.

Via Stock ETFs: Managed by fund managers selecting a diversified portfolio of stocks, balancing risk and return. Requires paying moderate transaction costs and management fees but offers diversified risk. Products like First Financial Global AI Robotics and Automation Industry Fund are options.

Tracking Index ETFs: Passively track AI-related indices, with the lowest trading costs, lower management fees, and the best risk diversification. Products like Taishin Global AI ETF (00851) and Yuan Dazhi Global AI ETF (00762) have sufficient liquidity.

It is recommended to adopt a dollar-cost averaging approach to enter, spreading out purchases to average costs and avoid short-term volatility risks.

Cold Thoughts on Investing in AI Concept Stocks

The long-term growth trend of the AI industry is certain, but this does not mean all AI concept stocks are worth holding long-term. History offers warnings—Cisco, the first internet equipment stock, reached a high of $82 during the 2000 dot-com bubble, then fell over 90%. After 20 years of good management, its stock price has yet to return to that high.

What does this tell us? Companies in the infrastructure layer benefit most initially, but high growth and market enthusiasm are unlikely to be sustained long-term. As the industry matures, these companies will eventually become stable but slow-growing.

Downstream application companies may seem more promising, but history is equally unkind. Microsoft and Google are top players, yet their stock prices also fell sharply after market peaks, and they have struggled to return to previous highs for years. Yahoo was once an internet leader but was eventually replaced by Google.

The key is to switch horses in time. This is not easy for ordinary investors. During phased investments, continuous attention to the speed of AI technology development, monetization capabilities, and whether individual stocks’ profit growth slows is essential.

Three Major Variables Affecting AI Concept Stocks

Macroe Liquidity: The interest rate policies of the Federal Reserve and other central banks are crucial. Loose policies favor high-valuation tech stocks, while high interest rates compress valuations. AI concept stocks are highly sensitive to interest rates and require close monitoring of policy trends.

Policy and Regulation: Governments worldwide generally view AI as a strategic industry and may increase subsidies or infrastructure investments. However, issues like data privacy, algorithm bias, copyright, and ethics could lead to stricter regulations. If regulation tightens, some AI companies’ valuations and business models will face challenges.

Market Sentiment: Although AI remains a focus, it can experience significant short-term volatility. The emergence of new themes like renewable energy and others may also divert funds.

Investment Outlook for AI from 2025 to 2030

The overall trend is “long-term bullish with short-term volatility.” In the short term, chip and hardware suppliers like NVIDIA, AMD, and TSMC will continue to benefit most. In the medium to long term, AI applications in healthcare, finance, manufacturing, autonomous driving, and retail will gradually land, translating into more actual revenue for enterprises.

A relatively stable strategy to participate in AI growth is to prioritize infrastructure providers such as chipmakers and server accelerators, or select specific application companies like cloud services, medical AI, and fintech. Diversified investment through AI-themed ETFs can effectively reduce the risk of individual stock price fluctuations.

For ordinary investors, adopting a long-term allocation and gradual entry, rather than chasing high in the short term, is the best way to reduce market volatility impact.

Several Risks to Recognize When Investing in AI Concept Stocks

Industry Uncertainty: Although AI has existed for decades, large-scale application is recent. Rapid changes and advancements make it difficult even for the most knowledgeable investors to keep pace. Investors may fall into sharp stock price swings driven by hype after purchase.

Unproven Companies: Many major tech firms are involved in AI, but numerous AI startups have little history or foundation for reference. These companies carry higher operational risks compared to well-established, time-tested firms.

AI-Related Risks: Leaders in computer science and related fields have warned of AI’s potential dangers. As the field expands and evolves, public opinion, regulations, and other factors may change unexpectedly, affecting the performance of AI concept stocks in unforeseen ways.