Andy Nurumov

Author: Andy Nurumov

October 2024 | San Francisco, US

I. Big Picture and Movements

Gist: The top five will dominate the App AI layer. The reason: they need to create hundreds of billions of USD in annual revenue for their investors, who invested at 50x (or more) revenue multiple valuations (and still do). They cannot achieve this quickly through LLM monetization alone.

Go Deeper:

After the last two years of intense investments in improvements to off-the-shelf LLM models, players outside the top five have already lost the game in the global market where they planned to compete. At the same time, do they earn credibility that could leverage brand and distribution to propose other products? Some—yes. Is it enough to win a new category? The answer is no. This is the time for them to update their product portfolio and the services they provide.

Meanwhile, some have had initial large or small wins and are trying to leverage them in adjacent segments. What is important is that almost all large players are starting apps for businesses or in B2C.

The question is: what is better for them to do now? Fortunately, there have been $10–500M fundraising rounds in these adjacent segments over the last two years—demand has already been tested for many segments. A status-quo strategy for them would be to play in the largest segments. Build or buy? The most innovative can build.

II. What Does the AI App Layer Bring New?

Gist: Business operations restructuring and new service models in established fields. Pricing and business models will evolve to create or meet demand.

Go Deeper:

In one of the latest Sequoia reports, they compare the cloud and mobile waves with the AI wave (see the link for details). Thinking in a similar framework:

• The Internet brought a new way of distribution, but the cloud made a new way of distribution for tech and media products possible. Pricing and business models evolved to earn more and meet COGS driver structures.
• Mobile brought new user experiences and ways of distribution. New business models leveraging attention emerged (social media became “TV” for Generation Z), leading to a significant transition of ads to mobile.
• Some incumbents have undergone cloud transformation, and many have transitioned through the mobile wave. Almost all large players will go through the AI wave.
• Moreover, many conservative industries, such as financial services and advertising, wanted to become tech companies. Now, they and others want to be AI companies.

What is important is that distribution and business models have been addressed. The current AI wave is different from merely changing distribution; it does not change (at least not yet).

Mainly, AI will change how business operations can be organized to “get the job done,” as well as user interaction with service providers.

At the same time, are B2C customers ready for a different experience of classical services? There should be (and will be) a change in pricing and business models for this. If providers offer the same prices, that is very unlikely. A 10x lower price or even a freemium model is more likely, particularly for customer segments that were previously unserviceable due to pricing.

New pricing and business models are not about reinventing the wheel. Ben Franklin used “SaaS pricing” for his library nearly 300 years ago. It is more about finding the right incentives.

Gist: New players should (a) focus on the reasoning component, (b) play in medium-sized niches, and leverage personalization. Established players should additionally (c) see this AI wave as a “continuous improvement activity” and leverage their partners’ R&D and teams.

Go Deeper:

The five largest players in AI will dominate every large domain where “generalization” can win over “specificity.” At the same time, this is not just another wave that software engineers can easily navigate, especially those without a background in physics or mathematics. You need to create the “brain” of your company; often, this means many “brains” and interaction between them. There is no library you can easily rely on—you need these “construction” abilities. Thinking of LLMs merely as “speaking heads” for creating a “set of brains” is limited; what is more important is understanding the fundamental problems that need to be solved for the transition.

Returning to competitive advantage, you can create a moat (at least temporarily) in two of the following ways:

Reasoning Layer: Below are examples where the reasoning layer wins over improvements in general models: (a) where wisdom wins over “knowing all” and (b) where creativity wins over “knowing all.”

Play in Medium-Sized Niches with Personalization Proposal: You can create the best products and leverage medium distribution to win in medium niches. By fine-tuning the “speaking head,” you can provide personalization. Companies have already created personalized CRM systems to replace their Salesforce subscriptions, etc. However, not all players can have an in-house team to do this; you could be such a provider.

(Remark: Personalization could become a general form of “competitive advantage” for AI-based companies, just as network effects became for marketplaces.)

As a Status Quo: Finally, leverage the R&D and teams of your partners. They (IT and consulting partners) would love to write you a check for this, and you would benefit from their support for your PR for the next two years. Given that the average laptop worker could eventually be replaced by an AI “speaking head” or “mouse/keyboard clicking” sooner or later, there will always be a business case and outsourcing opportunities to implement it. This is perfectly acceptable; very few can manage innovations like the DeepMind and Google combination does.

Copyright: Citation and paraphrasing usage are welcome with a hyperlink to the original article.

About the author: Andy Nurumov specializes in Fintech, AI, and Marketplaces. He is a Founder and ex-CEO of an 8-figure B2B E-Commerce marketplace for Enterprise clients, which he scaled to a team of 20+ people and positive cash flow.