OpenAI's SearchGPT: More than Google, Glean should take note
Enterprise search is $120Bn opportunity..and OpenAI has 12 months of runway left
3+1 things about OpenAI’s foray into search business:
OpenAI is piloting SearchGPT, its search engine.
OpenAI training and inference costs could reach $7bn for 2024!
OpenAI is set to lose $5bn this year - the company has 1,500 employees, makes $2Bn from ChatGPT and has runway of ONLY 12 months left.
(Bonus) OpenAI is red-eyed focus on enterprise customers - giving them more control and administrative tools to make ChatGPT Enterprise the de-facto infra play in enterprise AI space.
Essentially, the company is losing money and given how difficult is it to monetize (and retain) consumers, I’d like to believe that OpenAI will roll out Enterprise AI search in no time - for them, consumer search is a way to test how easy/difficult is it to work with a variety of data partners/publishers (with their own data formats) and apply the learnings in B2B.
Let me explain:
SearchGPT Pilot
OpenAI's pilot of SearchGPT signals a strategic move into the search engine market. This initiative likely serves multiple purposes:
Testing the waters for a consumer-facing search product
Gathering data and insights on search behavior and patterns
Developing and refining algorithms for information retrieval and relevance
OpenAI's Massive Training and Inference Costs
The projected $7 billion in training and inference costs for 2024 underscores the immense resources required to maintain and advance AI models at OpenAI's scale. This high-cost base necessitates:
Aggressive pursuit of revenue streams to offset expenses
Potential pressure to monetize products more quickly
Likely focus on high-margin enterprise solutions
Where is the runway?
With projected losses of $5 billion this year and only a 12-month runway (as per The Information), OpenAI is under considerable pressure to achieve profitability:
The $2 billion revenue from ChatGPT, while substantial, is insufficient to cover costs
1,500 employees represent a significant fixed cost
This financial situation likely accelerates the push into enterprise markets for revenue stability
Enterprise Focus
OpenAI's emphasis on enterprise customers through ChatGPT Enterprise indicates:
OpenAI has been giving enterprise customers a lot more control and administrative tools to make ChatGPT Enterprise the de-facto infra play in enterprise AI space.
B2B equates to higher margins and revenue stability
A case for (boring) enterprise search
Given these factors, it's highly possible that OpenAI will rapidly expand into enterprise search for several reasons:
Monetization Potential: Enterprise search offers a clearer path to monetization compared to consumer-facing products.
Leveraging Existing Technology: OpenAI can repurpose much of its SearchGPT technology for enterprise applications.
Data Partnerships: Experience gained from consumer search regarding data partnerships and publisher relationships can be applied to enterprise contexts (easily!).
Competitive Advantage: OpenAI's advanced language models could provide a significant edge in understanding and retrieving complex enterprise data.
Integration with Existing Products: An enterprise search offering could integrate seamlessly with ChatGPT Enterprise, creating a more comprehensive AI suite for businesses.
Enterprise search is $120Bn opportunity
If the share of software spend that shifts to GAI applications (Generative Artificial Intelligence) were to reach 60% in a decade and 10% of the global labor force were to use the applications, the total revenue from GAI applications would reach $118.6 billion. If 50% of software spend went to GAI but 15% of the global labor force were to use them, the total revenue in this scenario would climb to $148.2 billion (via)
As of now, Glean is a clear winner in this space with more than $200mn raised and is at $30mn ARR (and of course, Cohere is trying to become a big player in this space).
OpenAI by the virtue of its partnership with Microsoft and channel partners like PwC stands to win the enterprise AI game - if played well (they are hiring customer success managers across different geographies).
What's your take?
Sam, you listening?