Can you still chase the AI chip trend? A comprehensive scan of AI stocks in 2025

Since the advent of ChatGPT, AI concept stocks have experienced a wild surge. Many investors, seeing cases of stock prices doubling, are eager to jump in and share the gains. But the question is: Are AI stocks still worth buying? Which AI stocks have real fundamentals? How can one avoid chasing high and getting caught?

Current State of the AI Industry Chain: From Speculation to Rationality

From the second half of 2022 to now, the performance of AI concept stocks has been a “rollercoaster.” On one hand, industry demand is indeed growing rapidly; on the other hand, stock prices have already heavily discounted future expectations.

According to the latest IDC report, global enterprise spending on AI solutions and technologies is expected to reach $307 billion by 2025. By 2028, total AI expenditure is projected to surpass $632 billion, with a compound annual growth rate of about 29%. Among these, accelerated server spending will exceed 75% by 2028, becoming the core hardware supporting AI deployment.

What does this data indicate? The AI industry is still in a high-growth phase, but the growth logic is changing. Early investors relying on “buy chips to make money” models are gradually losing effectiveness. Now, more precise selection of segments within the AI industry chain with genuine growth momentum is necessary.

Many institutional investors have already adjusted their strategies. Bridgewater Fund increased holdings in key AI players like NVIDIA, Alphabet, and Microsoft in Q1 2025, reflecting continued optimism about the AI industry but with more cautious selection.

Three Major Profit Opportunities in the AI Industry Chain

The AI industry chain can be divided into three key segments: chips and hardware, cloud computing and infrastructure, applications and software services.

Upstream: Chip Suppliers Are the Most Popular

GPU chip manufacturers remain the core beneficiaries in the AI industry chain. NVIDIA, as the leader in GPUs, has a market cap of $4.28 trillion, with its stock price rising over 11 times since ChatGPT’s launch. This is no coincidence — it holds an absolute voice in the global AI acceleration chip market.

NVIDIA’s moat lies in: leading global computing power, a complete ecosystem, and a large developer base. Even with competitors, it’s difficult to shake its position in the short term. NVIDIA’s revenue grew over 120% in 2024, hitting new highs in Q2 2025, with net profit increasing over 200% annually.

However, it’s important to note that NVIDIA’s current valuation is already high (PE around 60), and market expectations are fully priced in. Whether future growth can continue to exceed expectations depends on whether generative AI applications truly achieve large-scale deployment as anticipated.

Network communication chipmaker Broadcom is also worth watching. Its market cap is $1.63 trillion, with FY2024 revenue of $31.9 billion, of which AI-related products account for rapidly rising to 25%. Broadcom’s logic is: AI development relies on high-speed networks; whether for chip interconnects or data centers, high-end network chips are essential. Broadcom has surged 3.51 times over the past two years, with target prices mostly above $2,000.

AMD, which can produce both CPUs and GPUs, is the third option. Although its GPU performance currently lags behind NVIDIA’s H100, the MI300 series offers a clear price advantage (about half the price of H100) and has begun to be adopted by major cloud service providers. AMD’s revenue in 2024 was $22.9 billion, with data center business growing 27% annually; in Q2 2025, it grew 18% year-over-year. If AMD can continue expanding its market share, there is still room for growth.

Midstream: Cloud Computing and Data Center Infrastructure

Microsoft and Google, though tech giants, have shown relatively stable performance in the AI concept space. Microsoft’s market cap is $3.78 trillion, with a 20.63% increase since the beginning of the year; Google’s market cap is $3.05 trillion, up 32.50%. Both companies have actual AI applications (like Copilot, Gemini) rather than just hype.

Downstream: Expanding Application Scenarios

AI applications in medical diagnostics, financial risk control, autonomous driving, and intelligent manufacturing are accelerating deployment. While these companies may not be as high-profile as upstream chipmakers, their long-term growth potential is more stable.

How to Invest in AI Stocks?

Direct Stock Picking vs. Buying ETFs

Choosing between directly buying AI concept stocks or thematic ETFs depends on your risk tolerance and investment style:

• Individual stocks: Higher risk but greater potential returns. Suitable for investors with research ability and ability to track industry trends. Lower transaction costs but require frequent portfolio adjustments.

• Thematic ETFs: Diversified risk, less management pressure. Suitable for those wanting to “buy a basket of AI industry stocks.” Fees are relatively low, but flexibility is limited.

As of Q1 2025, global AI and big data funds have assets exceeding $30 billion, indicating that such products have become mainstream investment tools.

Lump-sum Investment vs. Dollar-Cost Averaging

Given the volatility of AI stocks, it’s advisable to adopt a dollar-cost averaging strategy rather than chasing high with a lump-sum purchase. This approach averages costs and reduces the risk of being caught in a downturn. Especially in uncertain macro environments, DCA is particularly important.

Risks of AI Stocks Cannot Be Ignored

High Industry Uncertainty

Although AI technology has existed for years, large-scale commercialization of generative AI is still in its early stages. Rapid technological development makes it difficult for investors to accurately judge which companies will truly benefit. Stock prices can fluctuate significantly based on news, requiring strong risk tolerance.

Valuation Overextension

Many AI concept stocks already price in growth expectations for the next 3-5 years. If actual growth falls short, valuations may face correction. This underscores the importance of timing when buying — chasing high carries significant risk.

Policy and Regulatory Changes

While governments support AI development, regulations around data privacy, algorithm bias, and copyright are tightening. If regulations become stricter, some AI companies’ business models and valuations could be challenged.

Unproven Business Models

Although large tech firms are actively involved in AI, many startups and new entrants have limited track records, making their risks higher.

Investment Recommendations for AI Stocks (2025-2028)

Short-term (6-12 months)

Continue to favor upstream chip suppliers, especially those with high order visibility and sufficient capacity. Avoid blindly chasing highs; wait for appropriate correction points to build positions. Keep an eye on policy developments, as Federal Reserve rate policies may pressure overvalued tech stocks.

Medium-term (1-2 years)

Expand from pure “chip concept” to “AI application deployment.” Companies in healthcare, finance, manufacturing, and other sectors that are already generating actual revenue from AI applications will have more sustainable growth.

Long-term (3-5+ years)

Adopt a long-term allocation strategy, balancing investments between AI infrastructure and application solutions. Avoid short-term chasing; instead, buy in batches and adjust periodically, allowing time to realize gains.

Final Advice

The AI industry is indeed a major trend for the future, but not all AI stocks are worth buying. The key points are:

1. Identify genuine demand: Look at hard data like order volume and capacity utilization, not just hype.
2. Beware of valuation traps: High growth does not equal high returns; assess whether entry prices are reasonable.
3. Diversify risk: Don’t put all eggs in one basket.
4. Regularly review: Periodically check if your investment logic still holds and if market conditions have changed.

Overall, the AI stock boom has not fully passed, but the era of blindly chasing highs is over. Rational selection, cautious positioning, and long-term holding are the right strategies to harvest returns in this cycle.