Artificial intelligence (AI) has captivated the world's imagination. However, investors shouldn't anticipate significant financial benefits from AI services in 2024, with a few exceptions. According to Gil Luria, managing director at D. A. Davidson, most companies won't be able to directly monetize AI services. While businesses are increasing their use of AI, software providers are more likely to offer AI features as added benefits to their existing offerings or as incentives for upgrades rather than selling them as standalone products. Luria explains that while the tools built around generative AI are compelling, they are still in their early stages and working out some issues. Companies' inclination will be to bundle these offerings to drive adoption, thus making them less of a material growth driver at the moment. This perspective may surprise investors who have shown great interest in the AI trade. The software ETF IGV has seen a roughly 50% increase in the past year, surpassing the 34% gain for the Nasdaq Composite (^IXIC) and the 20% gain for the S&P 500 (^GSPC). Luria attributes this exceptional performance mostly to AI excitement. He believes that most software companies will face challenges if they do not deliver tangible AI results this year. However, Microsoft and Adobe are exceptions to this trend. D. A. Davidson considers them top picks for mega caps due to their advanced AI capabilities and solid businesses outside the field. Adobe, well-known for its text-to-image generation, has integrated generative AI features across its creative suites, including a generative AI credits system that incentivizes users to pay more to access those functions. Microsoft, with its Copilot services, charges extra for using the AI assistants embedded within Microsoft 365, GitHub, and other platforms. However, the primary source of AI revenue for Microsoft remains its cloud service, Azure, which offers access to large language models from OpenAI and other providers that developers can use to create their own applications. Even though monetization may not happen immediately, software that supports customer-facing roles is well-positioned for the future.

Generative AI has enormous potential, but current use cases mainly focus on tasks like text generation, text summarization, and text-to-image generation. Salesforce's AI assistant, Einstein Copilot, is set to launch publicly at the end of February and incorporates these functions, including pre-populating emails, summarizing calls, and answering common questions. Morgan Stanley recently upgraded Salesforce's stock to a top pick, citing its consistent revenue growth and relatively lower trading multiples compared to peers. In the age of AI, data is highly valuable, so investors should look for companies at the forefront of data access and integration. Salesforce, for instance, already possesses a wealth of data from businesses using its products for automating customer service, marketing, and other tasks. In December, the company announced a new capability to import unstructured data, such as emails and phone calls, into its Data Cloud product, which connects data from various software offerings. The idea is to allow AI assistants to integrate business context into prompts, providing more information for better AI performance. While large, universal knowledge models like ChatGPT attract attention, companies that pioneer more efficient models will be the ones reaping the financial benefits. For instance, workers at lamp makers do not require help with drafting emails about the Roman Empire. Smaller models that contain only relevant information would be cheaper to operate, allowing businesses to generate profits. Microsoft and Salesforce, among others, are already working on such initiatives. Citi analysts reported that Microsoft has formed an internal team to develop smaller AI models (SLMs) that are cheaper to run and reduce dependencies on OpenAI. These SLMs, like Phi-2 and Kosmos-2, aim to deliver performance comparable to larger models but with significantly fewer parameters. Looking ahead, one foreseeable future for enterprise AI is increased interoperability, where AI assistants can communicate and collaborate with each other. This would enable an AI assistant managing your calendar to cooperate with another handling client information. Johnson from Salesforce envisions a scenario where the assistants help each other to benefit end-users. If these AI assistants can generate profits for their respective companies, investors will stand to benefit as well.