Over the past few months, various software companies have either launched or announced their own AI copilot products, such as Microsoft's Github copilot, Office copilot, Windows copilot, Dynamics copilot, Salesforce's Einstein Copilot, and many others. Their features and pricing vary greatly. In order to assess whether their pricing is reasonable and to gauge potential price increases in the future, we have proposed the following scoring system for comparison. This article comes from BEDROCK member Jimmy, along with his team's discussion records.
Basic Analysis Framework: Firstly, we consider how much value this product brings to users.
This can be further broken down into three aspects:
Firstly, how much efficiency can the AI copilot help users improve, which determines how much time can be saved (each 20% improvement earns 1 point); Secondly, the user's salary level, which reflects the time cost saved for them (an average salary of $60,000 earns 3 points); Thirdly, the AI's other capabilities/advantages, which can expand the boundaries of human capabilities and do things that humans cannot do, such as batch processing tasks (thousands of events in the field of network security), and so on.
Based on this, here are the scores for some software companies:
Among them, Github, as a programmer's tool, has high scores in efficiency improvement, user salary, and AI's additional capabilities (debugging, reading others' code, etc.); Office has a slightly lower efficiency improvement score, and user salary and AI's additional capabilities (creating PPT based on Word, etc.) are average; Adobe has moderate efficiency improvement because it still requires a lot of manual operation, but user salary and AI's additional capabilities (batch image generation, providing inspiration, etc.) are relatively high; CRM, NOW, and Office are similar, with NOW having higher user salaries (IT professionals); CRWD and PANW, as IT security companies, have high scores in efficiency improvement, user salary, and AI's additional capabilities (handling a large number of security events, providing insights based on log data, etc.); Finally, SNPS, as an EDA (Electronic Design Automation) tool, has relatively lower efficiency improvement for users, but user salaries are very high, and it can propose some optimization methods that humans might not think of, which is highly valuable in chip production.
From the average scores, IT security companies CRWD and PANW have the highest potential value-add, followed by Github, and then ADBE and SNPS. Lastly, NOW, CRM, and Office. SNPS might be slightly underrated because its value primarily lies in the overall value enhancement in chip design rather than direct user value improvement, while others align with common sense.
On the other hand, consider how much value the company can capture from AI copilot.
Even if the value created by AI copilot is high, software companies can only capture a small portion if the competition is intense. We also consider three aspects: Firstly, the importance of this software tool to users, how mission-critical it is; Secondly, the level of competition in the AI copilot market, whether there are alternatives; Finally, whether it might trigger an arms race and involve all users.
In terms of mission-criticality, Office and NOW have lower scores, CRM scores slightly higher because it is a revenue-generating tool, and others like Github, ADBE score the highest. In terms of substitutability, CRWD has the lowest score because there is strong competition from MSFT Windows Defender in the endpoint protection field; SNPS has the highest score because it has very poor substitutability in its niche. Other software tools have varying degrees of substitutable alternatives. Regarding the arms race, IT security companies CRWD and PANW compete with hackers, making it the most intense arms race; EDA tool SNPS is also crucial for chip optimization; Github/ADBE/CRM have a large user base and are results-driven, ultimately forcing everyone to use AI to improve their efficiency; Office and NOW, as communication and internal management tools, have a slightly weaker level of arms race.
In the end, from the average scores, EDA company SNPS and IT security company PANW have the highest potential take rate, followed by ADBE and CRWD, then Github and CRM, and finally Office and NOW, which aligns with common sense.
Combine these two tables to view:
PANW has the highest score, followed by CRWD and SNPS, then Github and ADBE, and finally CRM, Office, and NOW.
On the other hand, from a more quantifiable perspective of how much they can charge, based on the publicly disclosed prices and guidance, the pricing of most products is currently relatively low, especially Github. ADBE's AI tools are more usage-based, so the recent price increases do not necessarily reflect the pricing of its AI products, so its take rate reference value is not high. In the long term, if we assume the long-term take rate is simply the average of the calculated take rate scores above multiplied by 2 divided by 100, then CRM and NOW have limited price increase potential, while other companies still have significant room for price increases. It's worth noting that the results for CRWD, PANW, and SNPS are not very meaningful because their product value lies not in improving user efficiency but in enhancing product capabilities (cybersecurity products, designed chips).
Of course, the above is just an analysis framework we have attempted to establish based on the current situation. The specific situation may change significantly over time in terms of user value addition and competitive landscape, so we need to maintain a cautious attitude and make adjustments as needed.
Comments