Summary of Key Points
Recently, the stock market for enterprise software companies in the U.S. has experienced extreme fluctuations within half a month, characterized by first soaring and then plummeting prices. Two weeks ago, leading software companies released their latest financial reports, which clearly demonstrated that AI has transformed from a costly expense into a new growth driver capable of generating billions of dollars in additional revenue. This led to a 14%-20% increase in the stock prices of companies like Salesforce and Snowflake. However, as soon as OpenAI announced the launch of Astra, an intelligent agent that can automatically perform entire processes across multiple software systems, the market was immediately pulled back by the old concerns about AI replacing traditional software, causing software stocks to fall by 3%-5%. Essentially, these fluctuations reflect the fact that the entire enterprise software industry, worth hundreds of billions of dollars, is being forced to completely overhaul its profit-making models that have been in place for over a decade. No one can yet determine whether AI will help software companies earn more money or whether it will completely disrupt the existing business models.
Detailed Analysis
1. Software Stocks Fluctuating Wildly in Half a Month: Investors’ Emotions as Intense as on a Roller Coaster
The sudden rise and fall in stock prices were a complete reversal of market expectations. From October last year to April this year, the S&P Software Index had already fallen by more than 33%. The main concern at that time was that AI would enable companies to write code and operate software on their own, meaning businesses would no longer need to purchase as many software products, leading to a mass sell-off of software stocks.
However, when the second-quarter financial reports were released, all the leading software companies provided concrete evidence of how AI could generate revenue, and investors realized their previous fears were unfounded. AI was actually helping companies earn more money, prompting them to buy back stocks and drive up their prices. But just two weeks after the price increase, OpenAI’s release of Astra sent everyone back into panic, as the issue seemed more urgent than ever. Astra directly threatened the traditional software model, leading to another sharp drop in stock prices.
2. Why Did the Market Become So Excited About Software Stocks Previously?
Previously, AI was seen as a costly expense for software companies, with billions of dollars spent on research and development. But the financial reports showed that AI could indeed generate revenue:
- The “Annual Recurring Revenue (ARR)” of software companies, which refers to the regular fees customers pay each year for updates, has seen a significant increase. For example, Salesforce’s AI-related products generated nearly $3.9 billion in annual recurring revenue, a year-on-year increase of 210%, which more than offset the fluctuations in traditional software licensing revenue.
- Snowflake, a company providing cloud data services, reported that its AI tools were used by over 9,100 enterprises, boosting its annual revenue expectations by $230 million.
- Even traditional software giants like Adobe and SAP saw their AI-related products generating over $500 million in annual recurring revenue, with 90% of their large contracts including AI services.
It’s like running a restaurant: you expected to spend millions on renovations, but after the renovation, customers were willing to pay twice as much for new AI-based services, bringing in more customers. Naturally, this led to higher stock prices for these companies.
3. Why Did Everyone Panic When Astra Was Released?
Astra represents a direct threat to the traditional software model. Previous AI products were merely tools embedded within software systems that required users to open specific applications to access them. Astra, on the other hand, acts as a fully autonomous assistant. You can simply instruct it to update customer records in the CRM system or send an email to the sales director, and it will handle all the necessary tasks without you even having to open any software.
Traditional software companies have relied on a model where they charge based on the number of users. If a company has 100 employees using their software, they charge a fee for each user. But with Astra, there’s no need for employees to use the software at all; you can simply assign tasks to Astra. Why buy 100 accounts when you only need one? By using one Astra account, you can save on 99 accounts, which directly threatens the traditional revenue model. Gartner predicts that by 2030, $234 billion in enterprise software revenue could be affected by this trend.
4. Software Companies Have Already Started Changing Their Pricing Models
In response to the threat of AI agents, leading software companies have begun to shift their revenue models. Instead of selling software licenses, they are now charging for the use of specific functions. For example, ServiceNow’s Action Fabric provides a unified interface for external AI agents, and Salesforce has introduced metrics to track the number of tasks completed by AI agents. These changes mean that software companies can charge based on the number of operations performed by AI agents. For instance, the platform’s AI agents have completed 7 billion tasks this quarter, with a 97% increase compared to the previous quarter.
This is similar to how cinemas used to rely on ticket sales. Now, cinemas can sell the rights to their content and charge based on the number of views, rather than relying on ticket sales.
5. It’s Too Early to Draw Conclusions, but the Rules of the Game Have Changed
The industry is still in a transitional phase, and no one can predict the ultimate outcome. It’s unclear whether the new revenue models based on per-use or per-task charging will make up for the losses from the traditional model. It’s also uncertain whether the cost of AI computing power will consume most of the additional revenue. What is clear, however, is that the competitive landscape for the software industry has changed significantly. In the past, companies with user-friendly interfaces and a wide range of functions were the winners. In the future, the most valuable companies will be those that control core business data, have established business processes over the years, and have robust compliance systems and permission management frameworks. Even the most advanced AI agents cannot simply alter financial data or bypass approval processes. The value of the software industry will be redistributed, and companies that rely solely on attractive interfaces and license sales will likely be the first to be eliminated.