#4 AGI Impact on Google, Microsoft, Salesforce, Amazon
Summary:
Alphabet is most exposed followed by Salesforce on a % revenue basis (in a max negative case)
Amazon would have second largest exposure on a $ basis (but last on a % revenue basis)
Max case is not the most probable outcome
Max Case Revenue at Risk by Company
max case explanation1
Details of table below by offering
Cloud computing
I assume, by and large, work will shift from humans to models. Under that case demand for compute (GPU) goes up across the board.
Search / Advertising -
Today State - Advertisers pay to get in front of people with purchase intent. This compensation to the ad network is -by and large- cost per click (CPC) times volume.
Future State - The Information stated Bing is introducing a version of OpenAI in its search results2 . The main question: How will this impact revenue per search?
The answer depends on the ROI that advertisers see. An ad to serve value to the advertiser when a certain % of users click the ad and make a purchase.
Upside case: The GPT like response is very informative and decreases the risk the potential buyer feels on making a purchase. As a result purchases increase relative to pure search. For example: I search for a desk chair that has a specific lumbar support. The current search and object information do not include this detailed information. In contrast, the inclusion of GPT does and I feel comfortable making the purchase.
Downside case:
Version #1 The GPT response answers my question in a way that is not truthful over time as a method to push product purchases. I begin moving away from ad funded search models.
Version #2 The GPT response answers my question truthfully. It says that the product won’t actually meet my needs and I do not purchase it.
Regardless there are some implications here for the valuation of Alphabet. Before revenue stability was assured. Now, rational investors will model downside scenarios. This grows the range of outcomes (risk). This risk results in larger discount rates. Larger discount rates result in lower present value of future cash flows. Finally, that results in a lower stock price.
Marketing Tech
Today State - Generally, sold on a per seat basis
Future State - This could go either way.
Upside Case - Advertising could be added into the models. This would be more complex for the advertiser to serve. This complexity could result in higher pricing.
Downside Case - Advertising in general becomes far less effective. As a result companies spend less on it. Due to this decrease there is less need for technology to manage it.
Service Tech
Today State - Generally sold on a per seat basis
Future State - This could go either way. I lean more toward the downside case.
Upside Case - Companies like Salesforce use the large language model APIs and offer them as a service via their install base.
Downside Case -
Version #1 Human agent seat sales decrease
Version #2 Customer Service becomes models interacting with models.
i.e. Customer A has their model continue to outreach until resolution occurs. Company X has their models respond to A w/ prior rules optimizing for LTV.
The current service tech companies3 have a very different type of engineer. This engineer is building workflow tools for human agents to use. In select cases it is building process builders for management to design.
This future state customer service technology company looks closer to a hedge fund. The decisions are probabilistic and real time.
The current service tech incumbents are not staffed to handle this. This is better created as a greenfield.
Conclusion
The max risk downside case would impact Google the most followed by Salesforce. This is not the most likely outcome but a starting point for analysis.
Why did I start w/ Max Case? I wanted to establish a boundary condition and work back from there. Revenue numbers are sourced from Seeking Alpha and are Last Twelve Months. Segment revenue numbers are sourced from most recent 10-Q from each company. In some cases I went into the financial powerpoint presentations to pull apart the segment earnings.
https://www.theinformation.com/articles/microsoft-and-openai-working-on-chatgpt-powered-bing-in-challenge-to-google
Salesforce, ServiceNow, SAP, Microsoft Dynamics, Zendesk, etc.