Summary in Plain Language
Recently, several seemingly unrelated events in the AI industry have come together: Salesforce, a global software giant that provides business services, saw its stock price rise by 22% on the day of its financial report release thanks to new AI-related products. Even OpenAI, the leading AI company, has stated that it will not get involved in developing specific applications to compete with other software companies, but will focus on building platforms instead. Domestic companies like Tencent, ByteDance, and Alibaba have also all entered the market to compete in the enterprise-level AI office solutions sector. These developments indicate that AI has moved beyond the easy phase of making quick money through writing copy, designing graphics, and coding, and has officially entered a more challenging phase where it needs to address the complex needs of traditional enterprises. The big model companies that were once in the spotlight are no longer the only players; platforms with enterprise customer relationships and mature software systems, as well as software service providers, have become the key players in determining the speed of AI adoption.
---
Detailed Analysis
1. Why are AI companies suddenly acting in unison?
The easy money to make has almost run out. In the past two years, AI has primarily targeted “sweet spots” with high tolerance for errors, such as writing copy, designing graphics, and generating code. Even if the initial AI-generated work had flaws, humans could easily make corrections. These tasks were already done by internet professionals who were receptive to AI, and all the necessary information was already available online, so the adoption speed was fast. However, this market is now approaching its ceiling. The remaining 90% of real business scenarios involve core operations of traditional enterprises, such as factory scheduling, financial payroll calculations, and supply chain inventory management, where mistakes can result in significant losses. Companies are no longer willing to risk using AI without careful testing. OpenAI’s acknowledgment that it had overestimated the speed of AI adoption reflects the end of this easy phase, forcing all players to compete in the more difficult areas that no one had previously tackled.
2. AI’s entry into traditional enterprises faces three major barriers:
- Unclear responsibilities: For example, if AI suggests a 20% decline in sales in a certain region, and upon further investigation, it turns out the actual decline was only 2%, no one wants to take the blame for such a mistake.
- AI alone is not enough: To avoid errors, AI must be integrated with the enterprise’s existing data systems and processes that have been verified over years. AI’s role is to translate the results into understandable language for humans.
- Different company rules: Each company has its own unique procedures, such as requiring departmental or managerial approvals before financial approvals. These custom rules must be tailored for each company, which is why there is a surge in demand for engineers who can provide customized implementations.
3. Software companies are now reaping the benefits of AI:
Previously, many feared that software companies would be replaced by AI. However, the situation has reversed. Software companies have a significant advantage: they have been serving customers for decades, understand industry processes, and have mature systems and loyal customers. AI companies lack these resources and cannot bypass them to directly serve traditional enterprises. Instead, AI has become a new driver of growth for software companies. For instance, Beisen, a domestic HR software provider, is seeing rapid growth due to AI-powered interview tools and training programs. It is predicted that AI-related revenue will become a major source of income for many listed software companies next year, debunking the notion of the “end of SaaS.”
4. Are software companies now competing with their own customers?
With the decreasing cost of AI, many company owners are considering developing their own AI solutions. While developing in-house systems used to be costly and ineffective, AI tools can now be used to create simpler, more efficient systems at a fraction of the cost. Software companies are opening up their systems and APIs, allowing customers to build their own AI applications while charging for usage, thus strengthening their relationships with them.
5. No single winner in the AI industry:
There will be no one-size-fits-all solution in the AI industry for a long time. On one hand, simple AI tools that don’t require deep integration with business processes are still useful. On the other hand, more complex AI applications that need to be deeply integrated with enterprise systems will be essential for managing complex operations. Both types of solutions will coexist, with the latter targeting much larger markets than the traditional internet sectors.