Can you still invest in AI stocks in 2025? This AI concept stock selection guide reveals the truth

Still debating whether AI concept stocks are worth entering? In 2024, AI-themed stocks have seen astonishing gains, but will this rally continue into 2025?

Actually, from the data, global corporate spending on AI is just beginning to explode. According to IDC forecasts, global AI expenditure will reach $307 billion in 2025, possibly soaring to $632 billion by 2028, with a compound annual growth rate of about 29%. This means the AI industry is far from reaching its ceiling—key is choosing the right stocks.

Three Investment Stages of AI Concept Stocks

This AI wave is similar to the internet boom, divided into three clear beneficiary stages:

First Wave: Hardware and Chip Suppliers

The earliest to profit are infrastructure providers. Companies like NVIDIA, AMD, and TSMC, involved in chips and accelerators, benefit from global data center upgrades. In 2024, NVIDIA’s revenue hit $60.9 billion, up 120% year-over-year—that’s the power of the first wave. In Q2 2025, NVIDIA’s revenue was over $28 billion in a single quarter, with net profit increasing over 200%.

Similarly, TSMC (2330), Taiwan’s largest market cap company, has overflowing AI chip foundry orders. AMD’s data center business is also accelerating, with a 27% growth last year and continued strong growth this year.

Second Wave: Midstream System and Solution Providers

Servers, cooling, and network switches start to show their value. Quanta (2382), through its subsidiary Quanta Cloud Technology, has become one of the world’s largest AI server OEMs, with 2024 revenue of NT$1.3 trillion, rapidly increasing AI server share. In Q2 2025, revenue exceeded NT$300 billion, up over 20% annually.

Also, Faraday Technology (3661), specializing in AI chip design services, posted full-year revenue of NT$68.2 billion in 2024, up over 50%. By Q2 2025, quarterly revenue surpassed NT$20 billion, doubling compared to last year.

Delta Electronics (2308), mainly providing cooling and power solutions, had annual revenue of NT$420 billion last year, with a rapid rise in data center business. As AI server power consumption continues to rise, cooling becomes a new bottleneck—an opportunity for companies like Zhaoxin (3324). This liquid cooling leader had revenue of NT$24.5 billion in 2024, up 30%. In 2025, major cloud service providers are accelerating the adoption of liquid cooling solutions, with Zhaoxin’s water-cooled module shipments soaring from Q2 onward.

Third Wave: True Beneficiaries at the Application Level

Microsoft (MSFT) is a typical example. By exclusive partnership with OpenAI, it deeply integrates Copilot into Windows, Office, Teams, and other products, covering over 1 billion users worldwide. In FY2024, revenue reached $211.2 billion, with Azure cloud services growing 28%, and AI services contributing more than half of the momentum. In Q1 2025, the intelligent cloud business revenue first exceeded $30 billion.

Google (GOOG), with a market cap of $3.05 trillion, is also pushing enterprise AI applications. MediaTek (2454), through its Dimensity chip series, is deploying in mobile AI and automotive AI fields.

Can AI Concept Stocks Be Held Long-Term?

That’s a tough question. Looking back at the internet bubble, Cisco (CSCO) hit $82 in 2000, then plummeted over 90% to $8.12 after the hype faded. Despite 20 years of solid operation, its stock price never returned to the high.

Similarly, Yahoo—once a internet giant—was eventually overtaken by emerging players like Google. This shows that even quality companies face huge risks if their technology path or market position change.

Therefore, upstream hardware and chip stocks (NVIDIA, TSMC, AMD, etc.) are more suitable for phased investments rather than long-term holds. High growth is often unsustainable; once market enthusiasm cools or new technologies emerge, valuations can sharply revert.

Downstream application stocks (Microsoft, Google) might be more stable theoretically, but still require regular review—if new competitors appear, market positions could be shaken.

Core advice: Instead of blindly holding long-term, adopt a phased investment strategy, judging when to exit based on AI technology development speed, monetization ability, and individual stock profit growth.

Three Main Drivers of AI Concept Stocks from 2025-2028

1. Policy Support

Governments worldwide are increasingly listing AI as a strategic industry. The US, China, and the EU are boosting subsidies and infrastructure investments, providing long-term positive support for the entire industry.

2. Accelerated Application Deployment

Medical diagnostics, fintech, autonomous driving, smart manufacturing—these are no longer concepts but gradually entering commercial use. Actual revenue conversion at the enterprise level will drive overall growth.

3. Continuous Explosion in Computing Power Demand

Advances in large language models and multimodal AI increase demand for servers, data centers, and cloud platforms, which remains unmet. IDC forecasts that by 2028, spending on accelerated servers will account for over 75%, becoming core to AI infrastructure.

But also be aware of risks:

• Valuation Fluctuations: Once the Fed shifts to rate hikes, high-valuation tech stocks are vulnerable to sharp declines.
• News Sensitivity: AI concept stocks react overly to news, leading to short-term surges and crashes.
• Capital Flows: New energy and other emerging themes may divert funds from AI.
• Policy Changes: Tightening data privacy, copyright, and ethical regulations could challenge certain business models.

How to Invest in AI Concept Stocks Most Efficiently?

Direct stock picking carries concentrated risk but has the lowest transaction costs. If you lack time for research, consider:

1. AI-themed Funds

Fund managers select various stocks to balance risk and return. The downside is higher management fees.

2. AI-related ETFs

Track indices, with the lowest trading costs, but may experience premiums or discounts. Examples include Taishin Global AI ETF (00851) and Yuan Da Global AI ETF (00762).

3. Dollar-Cost Averaging

Whether buying stocks, funds, or ETFs, use regular fixed investments to average costs and avoid chasing highs. Remember, even hedge fund holdings change, so continuous adaptation is necessary.

4. Platform Selection

Open accounts directly with brokers for Taiwan stocks, or via omnibus accounts or overseas brokers for US stocks. For short-term trading with leverage, contracts for difference (CFDs) platforms with no commission and higher leverage are options.

The Correct Approach to AI Concept Stock Deployment in 2025

Short-term strategy: Focus on chip and hardware suppliers like NVIDIA, AMD, TSMC—these are the most direct beneficiaries. But set profit-taking points and don’t hold too tightly.

Mid-term strategy: Quanta, Faraday, Delta Electronics—midstream system providers expected to benefit from server and data center construction peaks.

Long-term strategy: Application companies like Microsoft and Google—though AI positives may already be priced in, ongoing enterprise AI deployment can provide support.

The safest approach: Avoid all-in on a single stock; diversify through AI-themed ETFs. This way, you participate in AI growth while reducing risks from individual stock volatility.

Final words: AI concept stocks in 2025 still have opportunities, but it’s no longer a time for blindly chasing highs. Careful stock selection, phased deployment, and regular review are the keys.