Over the past two years, the biggest winner in AI has been GPUs.
In the next two years, the biggest winner might not still be GPUs.
Because what companies are truly willing to pay for has never been conversation.
It's efficiency.
The boss doesn't care how many questions AI can answer.
The boss is interested in:
Can we hire fewer people?
Can we make fewer mistakes?
Can we turn the work that 10 people used to do into just 3 people?
From this angle, ServiceNow (NOW) might be one of the most underrated software companies in the AI era.
---
## 1. A lot of folks actually don't get what NOW is selling
Many investors know about Microsoft, Salesforce, and Snowflake.
But it's unclear what exactly ServiceNow does.
In fact, it's not selling software.
But rather the efficiency of the enterprise.
Employees need to apply for permissions when onboarding.
If the computer breaks, it needs repair.
Finance needs to approve reimbursements.
Customer complaints need to be handled.
Security incidents need a response.
A large enterprise can have hundreds of thousands of similar processes happening every day.
In the past, coordination relied on emails, phone calls, and manual efforts.
The efficiency is extremely low.
What ServiceNow does is move all these processes to a unified platform.
Simply put:
Enterprises are responsible for making money.
ServiceNow is responsible for making enterprises run smoothly.
---
## II. The biggest misconception in the market: AI will replace enterprise software
I think that's wrong.
ChatGPT can tell you how to reimburse.
But it can't complete your corporate reimbursement process.
Claude can help you write work orders.
But it can't execute tasks across systems.
Big models are responsible for thinking.
Enterprise software is responsible for execution.
These two are not in competition.
Instead, they empower each other.
In the future, every company will have more and more AI Agents.
The real question is:
Who will manage these Agents?
Who will assign the tasks?
Who will control permissions?
Who will connect ERP, databases, and internal systems?
The answer won't be big models.
And it's more likely to be a workflow platform like ServiceNow.
The more AI spreads.
The workflow platform becomes even more important.
---
## III. What's NOW's biggest moat?
Not the brand.
It's not AI.
But rather the switching costs.
Deploying ServiceNow in a large enterprise often takes years.
IT systems, human resources, financial processes, and security management are all deeply intertwined.
Switching a chatbot is easy.
Replacing the core workflow platform of an enterprise is extremely difficult.
Because this means:
Rebuild processes.
Retrain employees.
Reconnecting systems.
Assume the risk of business interruptions.
That's also why ServiceNow maintains an extremely high customer retention rate.
Many customers have been using it for over ten years.
---
## IV. AI is becoming NOW's second growth curve
In the past, the market valued NOW mainly based on traditional enterprise software.
Now the situation is changing.
More and more companies are starting to experiment with AI Agents.
But companies quickly realize:
The real challenge isn't the model.
But rather management.
What can the Agent access?
What systems can the Agent call?
What tasks can the Agent perform?
What happens if the Agent makes an error?
These issues ultimately point to the workflow platform.
ServiceNow has launched products like Now Assist and AI Agent management platform.
Essentially, it's a battle for the position of the enterprise operating system in the AI era.
If in the future every company has dozens or even hundreds of Agents.
Then the platform that manages these Agents will also be revalued.
---
## V. Why I started to seriously pay attention to NOW
In the past two years.
Market funds are mainly concentrated in GPUs, storage, servers, and data centers.
This is the first phase of AI.
Sell the shovels.
And I believe we are entering the second phase of AI.
Companies start to think:
How to really utilize AI.
How to improve efficiency.
How to reduce costs.
How to make AI do the work for employees.
At this time, the importance of software will be reassessed by the market.
If GPUs are the most important infrastructure of the AI era.
So, the workflow platform is likely to be one of the most important toll booths in the AI era.
Big models are responsible for answering questions.
ServiceNow is responsible for getting the work done.
What companies are ultimately willing to pay for is always the latter.
Buy the moat, ride the bull — buy the moat, hold the bull.
