作者:Investor's Business Daily
The big run of artificial intelligence stocks has taken a breather in 2025. While semiconductor stocks ruled as the best AI stocks early on, there could be new plays emerging. To be sure, top AI stocks such as Microsoft (MSFT), Nvidia (NVDA) and Palantir Technologies (PLTR) face high expectations.
For many companies — such as Google parent Alphabet (GOOGL), Amazon.com (AMZN) and Facebook parent Meta Platforms (META) — the rise of generative AI poses both risk and opportunity.
Amid the rise of generative AI — which generates text, images, and video — it's a good time to be cautious amid the hype, especially given recent developments at Super Micro Computer (SMCI).
↑ X NOW PLAYING The AI Boom Is Kicking Off A Tech Supercycle. Futurist Amy Webb Predicts Even More Radical Transformation Ahead.
Many companies suddenly tout AI product roadmaps. In general, look for AI stocks that use artificial intelligence to improve products or gain a strategic edge.
What's more, data analytics software maker Palantir reports first quarter earnings on May 5. Palantir recently hosted its sixth AI event. Palantir stock has gained 64% in 2025 after soaring 340% last year.
While software companies have been slow to monetize AI products, some analysts still expect long-term gains.
"We believe the worst is behind us for software fundamentals with leading growth indicators stabilizing," said Goldman Sachs analyst Kash Rangan in a report. "We see AI as the sector's key catalyst, especially as focus moves from infra/training capital spending to inferencing use cases at the platform and application layers."
Among data center infrastructure plays, Arista Networks (ANET) reports Q1 earnings on May 6.
Meanwhile, Meta stock has gained 2% in 2025. On Meta's Q1 earnings call, CEO Mark Zuckerberg laid out five pillars of expected AI growth. They include improved advertising, engaging social media experiences, business messaging, the Meta AI app, and AI devices, including spatial computing.
The social networking giant in April launched the Meta AI app, built with its Llama 4 training model, with chatbot and web-searching features. Previously, Llama had been embedded in Meta applications such as Instagram and WhatsApp.
Amid worries over AI competition, Google stock is down 13% in 2025. But Q1 earnings provided some upbeat news for Google stock on the AI front. What's more, Google could be near a licensing deal with Apple (APPL) for its artificial intelligence engine Gemini.
Meanwhile, revenue growth at cloud computing giants has been a big theme. That's because capital spending on AI infrastructure has boomed at the cloud giants.
Google's cloud unit reported revenue growth of 28% on April 24, in-line with estimates. But Microsoft crushed Wall Street's fiscal third quarter targets amid strong sales of cloud computing and artificial intelligence services. Azure cloud revenue growth accelerated to 35% in constant currency, up from 31% in the prior quarter.
Microsoft is the biggest investor in generative AI leader OpenAI, having spent some $14 billion on the startup. It cloud business benefits from handling OpenAI's ChatGPT services.
Meanwhile, Amazon's cloud revenue growth decelerated to 17% from 19% in the prior quarter.
"Cloud growth can be lumpy, and while AWS is not seeing the benefit of ChatGPT usage like Azure, corporate spend remains steady," said Bank of America analyst Justin Post in a report.
Further, Microsoft's results boosted Nvidia-backed CoreWeave (CRWV), which recently launched an initial public offering. CoreWeave is a new AI cloud services provider that rents out Nvidia GPU-equipped servers.
Meanwhile, Nvidia stock is now down 15% in 2025.
A bellwether for AI stocks, chip maker Nvidia sold off after the Trump administration effectively barred the company from selling its custom artificial intelligence processors to customers in China.
Meanwhile, Nvidia hosted its GTC Conference in mid-March. For Nvidia, margin pressure has been a key issue as it ramps up production of next-generation Blackwell AI chips in 2025. At GTC, Nvidia also updated its AI accelerator roadmap. However, whether customers will really need "Rubin Ultra" GPUs in 2027 and "Feynman" GPUs in 2028 remains a big question.
Export rules are a wild card for AI chip makers.
For example, the Framework for Artificial Intelligence Diffusion categorizes countries into three tiers, affecting their ability to obtain AI semiconductors. While Trump administration discussions are still ongoing, one change "involves scrapping the existing system to allow U.S.-designed chips to become a more strategic tool in international trade negotiations," said a UBS report.
Meanwhile, ChatGPT-maker OpenAI was valued at $300 billion as part of a new $40 billion fundraising round led by SoftBank. OpenAI builds large, multimodal foundation models. The OpenAI funding round was good news for infrastructure AI stocks because OpenAI has plenty capital for future data center build-outs.
Investors should keep a close watch on the fierce competition in AI models. One view is that competition between U.S. and open source AI models from DeepSeek and others would heat up amid a U.S.-China trade war.
In a post on April 4, OpenAI's chief executive Sam Altman noted plans to release GPT-5 in "a few months." He added that it is much better than "we originally thought."
Meta on April 5 released its open source Llama 4 AI model family. Meta said the most powerful Llama 4 model, Llama 4 Behemoth, is still in training.
Google recently released Gemini 2.5 while Anthropic continues to upgrade its Claude family of AI models. Then, there's Elon Musk' xAI startup with its Grok models.
Safe Superintelligence Inc., co-founded by former OpenAI chief scientist Ilya Sutskever, has raised $2 billion at a valuation of $32 billion, said a report. Greenoaks, Lightspeed Venture Partners and Andreessen Horowitz took part in the funding round.
Meanwhile, China-based DeepSeek is readying the next version of its controversial open-source models. Generally, the AI models are battling in reasoning capabilities and other performance yardsticks as well as computing power needs.
Here's IBD's take on the key questions involving DeepSeek, including U.S./China competition in artificial intelligence. Meanwhile, Alibaba Holdings (BABA) has its Qwen AI models and Baidu (BIDU) has Ernie AI models.
Having struggled to generate new revenue from "copilots," software companies are now turning to autonomous, goal-driven AI agents. One big issue for software companies is how fast customers ramp up pilot programs to commercial deployment.
Palantir — as well as Snowflake and privately held Databricks — are focused on helping companies use proprietary data to build their own AI models. Here's a look at Databrick's strategy.
In an IBD interview, ServiceNow (NOW) Chief executive Bill McDermott explained about how the enterprise software maker aims to be an AI winner.
Meanwhile, software maker Salesforce (CRM) reported Q4 earnings that disappointed. The software maker's fiscal 2026 outlook came in light even though it's ramping up new AI agents.
The commoditization of AI models could spur application development. While "training" AI models has been the biggest driver of capital spending, the market will shift to "inferencing," or running AI applications, in the long run.
Among AI stocks to watch, Apple (AAPL) has lagged in 2025. Apple stock has shed 18%.
Apple Intelligence features in iPhone 16 models have not sparked a big upgrade cycle. With iPhone 17 models expected to debut in September 2025, it remains to be seen whether Apple Intelligence features are improved. Voice assistant Siri has yet to be upgraded with advanced AI technology.
Meanwhile, AI chip export restrictions remain a wild card on Nvidia stock as well as Advanced Micro Devices (AMD).
On the earnings front, Broadcom (AVGO) reported fiscal Q1 results on March 6. Broadcom stock popped on its AI chip demand outlook. Broadcom makes custom AI chips for cloud computing titans.
Qualcomm (QCOM), ARM Holdings (ARM), and Marvell Technologies (MRVL) are other AI chip makers to watch.
Also, AI technology uses computer algorithms. The software programs aim to mimic the human ability to learn, interpret patterns and make predictions.
Until recently, machine learning was largely limited to models that processed data to make predictions. The AI models focused on pattern recognition from existing data. Corporate spending on AI projects was modest as companies mulled return on investment.
Company | Symbol | Comp Rating | Industry name | AI angle |
---|---|---|---|---|
Nvidia | (NVDA) | 84 | Elec-Semiconductor Fabless | Cloud computing giants buying more chips to train AI models or run AI workloads. Big lead over rival Advanced Micro Devices (AMD). |
CrowdStrike | (CRWD) | 98 | Computer Software-Security | AI chatbots expected to automate more functions in security-operations centers and reduce the time to detect computer hacking. |
Arista Networks | (ANET) | 90 | Computer-Networking | Sells computer network switches that speed up communications among racks of computer servers packed into "hyperscale" data centers. With AI growth, internet data centers will need more network bandwidth. |
Microsoft | (MSFT) | 91 | Computer Software-Desktop | Biggest investor in generative AI startup Open AI, whose ChatGPT users require Azure cloud services. Microsoft's business AI assistant, Office 365 Copilot, is another potential revenue source. |
Salesforce | (CRM) | 73 | Computer Software-Enterprise | Pivoted to autonomous, goal-driven AI agents from conversational co-pilots. Expected to use a mix of subscription and consumption-based pricing. |
Amazon.com | (AMZN) | 80 | Retail-Internet | Alexa smart assistant upgraded. Cloud computing unit working with OpenAI rivals Anthropic, Hugging Face and Falcon 40B. |
New generative AI models process "prompts," such as internet search queries, that describe what a user wants to get. Generative AI technologies create text, images, video and computer programming code on their own.
Companies will aim to boost productivity by developing customized AI for specific industries. Proprietary company data will be used to train AI models.
AI systems require massive computing power to find patterns and make inferences from large quantities of data. So the race is on to build AI chips for data centers, self-driving cars, robotics, smartphones, drones and other devices.
For chipmakers, analysts expect a market for "edge AI" — on-device processing of AI apps to emerge.
What's more, one key question for investors is whether tech industry incumbents will be the big generative AI winners. Or, will a new wave of AI startups eventually dominate? OpenAI has told employees its now on an annual revenue run-rate of $3.4 billion, up from $2 billion in January.
Further, OpenAI recently laid out more details of its plans to adopt a for-profit business structure in 2025.
Large language models provide the building blocks to develop applications. Further, LLMs help AI systems understand the way that humans write and speak. Also, LLMs require training data for specific tasks. Companies with access to troves of data hold an edge.
OpenAI is part of a wave of LLM startups that includes AI21 Labs, Anthropic and Cohere. Anthropic's latest funding round values it at $61.5 billion.
Follow Reinhardt Krause on X, formerly Twitter, @reinhardtk_tech for updates on artificial intelligence, cybersecurity and cloud computing.
YOU MAY ALSO LIKE:
Want To Trade Options? Here Are The Basics To Get You Started
IBD Digital: Unlock IBD's Premium Stock Lists, Tools And Analysis Today
Learn How To Time The Market With IBD's ETF Market Strategy
Monitor IBD's "Breaking Out Today" List For Companies Hitting New Buy Points