First AI Fruits Will Come From AI-Enabling Existing Products

Capital investment at Alphabet, Microsoft, Apple, Amazon and Meta is up, reaching about $55 billion in the second quarter of 2024, with the expansion largely driven by investments in artificial intelligence capabilities. 

Some of that investment comes as the firms create their own generative artificial intelligence models and train them; other costs to adapt models to the existing core firm products. So the effort initially involves some mix of creating general-purpose models other enterprises can use and applying GenAI to existing products and processes.

Alphabet and Meta have been most active creating general-purpose models others can use. But all the hyperscalers are adding GenAI to their core revenue-driving products. 

For Alphabet that means grafting AI onto search, Pixel phones and the Android operating system. Microsoft initially focused on AI-enabling its productivity suite. Virtually all the firms are creating AI assistants. 

Meta is AI-enabling its social media and communication apps. Apple largely is AI-enhancing its iPhone apps and functions, while enhancing Siri. 

Amazon is applying GenAI to its shopping experience while hosting a number of leading models on Amazon Web Services. 

source: Bloomberg 

There already are clear examples of use cases across core shopping, social media, search, recommendations engine and communications and productivity suite use cases. But much more is expected. 

Still, it is worth noting that “the things that will drive the most results in 2025 and 2026 are…the ways that AI is shaping the existing products,” said Mark Zuckerberg, Meta CEO. That noted, the Meta chief also said that “we don't expect our Gen AI products to be a meaningful driver of revenue in 2024.”

Much of the value of recommendations across media consumption and shopping for example, hinges on the value (accuracy, relevance)  of the recommendations, as well as the ease of interacting with the apps. 

“Over time I'd like to see us move towards a single unified recommendation system that powers all of the content including things like people you may know, across all of our surfaces,” said Zuckerberg. 

For advertising use cases, AI should automate nearly the entire process. “Over the long term, advertisers will basically just be able to tell us a business objective and a budget, and we're going to go do the rest for them,” said Zuckerberg. 

In other cases, where general-purpose models are created, the effort is “enabling lots of people to create their own AIs,” said Zuckerberg. 

For the most part, all the AI efforts include creating ways to add AI features and functions to all existing products.

AI Gold Rush Exists Because of the "Rule of Three"

The present generative artificial intelligence "gold rush" exists for a good reason: the Rule of Three. In most instances, a stable competitive market never has more than three significant competitors, the largest of which has no more than four times the market share of the smallest,” BCG founder Bruce Henderson said in 1976. 

The caveat is that the rule does not work as well to industries that are unstable or heavily regulated, such as investment banking; consumer electronics; some parts of the IT software and services business; life insurance and parts of the telecommunications industry.

Industry	Top Three Players (Approximate Market Share)	Rule of Three Applies?
Automotive	Toyota, Volkswagen, General Motors (~30%, ~12%, ~10%)	Generally applies
Smartphones	Apple, Samsung, Xiaomi (~40%, ~25%, ~15%)	Generally applies
Soft Drinks	Coca-Cola, PepsiCo, The Coca-Cola Company (~40%, ~25%, ~15%)	Strongly applies
Airlines	Delta Air Lines, United Airlines, American Airlines (~15%, ~14%, ~13%)	Applies with nuances due to consolidation
Fast Food	McDonald's, Yum! Brands (KFC, Pizza Hut, Taco Bell), Restaurant Brands International (Burger King, Tim Hortons) (~15%, ~14%, ~12%)	Does not apply
Cloud computing as a service	Amazon Web Services, Microsoft Azure, Google Cloud Platform (~30%, ~20%, ~15%)	Generally applies
Retail (e.g., Online Fashion)	Amazon, Inditex (Zara), H&M (~40%, ~15%, ~10%)	Generally applies

Sometimes known as “the rule of three,”  he argued that stable and competitive industries will have no more than three significant competitors, with market share ratios around 4:2:1.

So one has to assume the same pattern will emerge for frontier GenAI models as well.

In 2023 alone, some 123 artificial intelligence foundation models, the building blocks of many modern AI applications, were released. In 2024, there may well be thousands of models in use, including both the smaller number of "foundation models" that lead the market, a larger number of general-purpose generative AI models that might be important in verticals, plus the many thousands of models that have been customized for use by specific enterprises.  

Virtually nobody believes all the would-be foundation models will survive, long term. And that fuels the "gold rush" mentality: only a few foundation models are likely to emerge as eventual market leaders, as tends to be true in any market.

While application markets tend to exhibit more diversity over the long term, compared to operating systems or semiconductor chip ecosystems, a reasonable argument can be made that, over the long term, just a handful of leading foundation models will lead the market, as that is the pattern in computing in specific and almost all markets generally.

Among the dozens of large foundation models that seem to be most used are the GPT series (OpenAI); the Claude series (Anthropic); PaLM and Gemini series (Google) as well as the LLaMA series (Meta). But there also are many small language models developing that generally are designed for specific purposes. 

In healthcare, SLMs might be used for medical document analysis, patient record summarization or perhaps research. In finance, SLMs might be used for fraud detection, sentiment analysis of financial news or risk assessment.

For customer service, SLMS might be used for chatbots. The point is that where LLMs were previously required, in the future SLMs might suffice. 

SLMs will be favored because they cost less. Training and deployment are more affordable, for example, since the models do not have to be trained on much-larger datasets. In some cases, SLMs can be developed faster. 

   

SLMs will be favored for industry-specific use cases, as well.  

Because there is less processing, there also is less energy consumption. Some argue SLMs also enable more privacy. 

Some Execs Say Generative AI Already is Boosting Revenues, Albeit at a Cost

According to a survey by Gartner, respondents have reported 15.8 percent revenue increases, 15.2 percent cost savings and 22.6 percent productivity improvement, on average, after deloying generative artifficial intelligence.

Gartner notes that GenAI requires a higher tolerance for indirect, future financial investment criteria versus immediate return on investment (ROI). 

One suspects we should take those quantifiable results with a bit of skepticism, as most of the returns from GenAI are indirect and hard to measure. 

Nevertheless, some argue that early adopters already are seeing revenue upside. A global survey of mid-market and enterprise firms conducted on behalf of Google Cloud suggests that 74 percent of organizations surveyed are currently seeing return on investment from their generative artificial intelligence investments.

Furthermore, 86 percent of respondents with GenAI in production mode claim annual revenues have climbed about six percent as a result. 

As with any survey of respondent attitudes, there is room for disagreement. Respondents might simply be inferring AI-driven growth when other forces are at work. Since many of the reported use cases deal with operations, contributions to revenue might often be estimates. 

Also, it might be the case that top-performing firms are most likely to be putting GenAI into use at scale. In other words, the top performers grow revenues more effectively, as a rule, and might be able to deploy new technologies more effectively as well. 

source: Google Cloud 

What we can probably say is that some firms supplying infrastructure, such as Nvidia, and some firms offering AI consulting, already can claim revenue boosts from AI. Accenture, for example, says its AI revenues for the first six months of 2024 were $2 billion. 

Boston Consulting group is projecting 20 percent of its 2024 revenue, and 40 percent of its 2026 revenue, will come from AI integration projects. IBM’s consulting arm has also made more than $1 billion from generative AI from WatsonX and generative AI, since inception. 

There’s a reason increasing use of generative and other forms of artificial intelligence is linked to data center capacity: model training is getting more compute intensive. So large language model training costs are growing. 

Still, generative AI costs are significant, both to create models and train them.

 

source: Epoch AI

A Gartner survey of 822 business leaders, conducted between September and November 2023, suggests that various generative AI projects cost between $5 million to $20 million. But that might not be the biggest impact, as costs for inference operations (asking questions, getting answers) could run between $8,000 to $21,000 per user. 

For a 1,000-user firm, that might suggest $8 million to $21 million annually in inference operations. 

source: Gartner 

All that noted, it might also be the case that some industries and use cases are more likely to be able to create direct revenue, though virtually any industry might claim indirect revenue benefits from any form of AI. 

Industry	AI Use Case	Direct Revenue Potential	Indirect Revenue Potential
Automotive	Autonomous driving	High (ride-sharing revenue, vehicle sales)	High (enhanced safety, driver experience)
Healthcare	Medical image analysis	Medium to High (e.g., diagnostic fees)	High (improved patient outcomes, operational efficiency)
Finance	Fraud detection	Medium (fraud prevention savings)	High (customer trust, regulatory compliance)
Media & Entertainment	Content generation (e.g., scripts, music)	Medium (licensing fees, content sales)	High (increased audience engagement)
Education	Personalized learning	Low	High (improved student outcomes)
Agriculture	Crop yield prediction	Low to Medium (potential premium for high-yield crops)	High (increased crop productivity, resource optimization)
Manufacturing	Predictive maintenance	Low	High (reduced downtime, increased efficiency)
Retail	Personalized product recommendations	Low	High

Some of us would not be at all surprised if disappointment with GenAI outcomes becomes more pronounced as projects seem not to provide the anticipated financial outcomes, in the near term. 

To the extent AI is the next general-purpose technology, as was the internet, we could ask the same questions about near term return from internet investments. 

How many firms will see near-term and quantifiable results from their capital investments and operating expenses directly related to GenAI? Perhaps anot so many.

Are AI Firm Valuations Really in a Bubble?

Many are concerned about “bubble” valuations for firms linked to artificial intelligence. And looking at price-to-sales ratios, Amazon, Alphabet, Meta, Microsoft and Nvidia have higher ratios than prior to the launch of ChatGPT, for example. 

That also applies to the “technology-heavy” Nasdaq index as well. Hence the concern about valuations being in a "bubble."

source: Blake Heimann, Wisdomtree 

But it might also be worth noting that price/sales ratios for Amazon, Alphabet and Meta are not out of line with their respective 10-year P/S ratios. In fact, Meta seems to have lower P/S than it did a decade ago. 

That cannot be said for Microsoft and Nvidia, both of which have much-higher P/S ratios than has been the case for most of the past decade. The Nasdaq index likewise seems to show significantly-higher P/S ratios at present than over the past 10 years. 

So if you are looking for possible overvaluation, Nvidia is where the greatest danger lies, with Microsoft and the Nasdaq index both showing possible overvaluation (or simply higher growth expectations). Right now we simply cannot tell whether the valuations are merited or not. 

source: Wisdomtree 

In fact, some might argue that Amazon has lower P/S and price-to-equity ratios than it has had over the last decade. Some will argue that is because so much of Amazon’s revenue comes from e-commerce and is at risk if a recession occurs and shoppers reduce their spending. 

Company	Current P/E Ratio	10-Year Average P/E	Current P/S Ratio	10-Year Average P/S
Alphabet (GOOGL)	22.5	19.8	5.2	4.7
Amazon (AMZN)	34.2	62.1	3.8	5.1
Meta (META)	25.8	22.3	8.1	7.4

As always, one’s perspective matters. Are sales going to grow enough to justify the higher ratios and valuations, or is too much growth expected? Or, looking at it another way, even if the higher growth does materialize, will that growth take longer than presently expected?

Which Generative AI platforms do Software Developers Use Most?

A Stackoverflow survey of developers including 65,437 responses from 185 countries suggests that ChatGPT is the generative artificial intelligence tool most used by software developers, followed by GitHub Copilot and then Alphabet’s Gemini. 

source: Stackoverflow 

AI use is increasing in most industries, say McKinsey consultants. In addition to general-purpose platforms such as OpenAI’s GPT-4; Alphabet’s Gemini or Meta’ Llama, industry-specific platforms are proliferating as well. 

Adobe Firefly is integrated into Adobe's Creative Suite, it excels in image and video generation for marketing materials and might be used by marketing and advertising staffs. 

Midjourney seems popular for generating highly artistic and creative images, especially in fashion and design.  

 

In healthcare, Nvidia Clara focuses on medical imaging, drug discovery, and genomics, providing specialized tools for healthcare professionals.   

Google DeepMind's AlphaFold is used in protein structure prediction, aiding in drug development and biological research.

BloombergGPT is tailored for financial data analysis and generation of financial reports, news, and code, and is used in the finance industry. 

Goldman Sachs's SECURE-LLM is built for secure financial applications, focusing on data privacy and compliance, and likewise is used in the financial industry. 

Nvidia’s  Drive Sim is used in the automotive industry for simulating autonomous vehicle environments and training AI models for self-driving cars.

   

Tesla's Autopilot is used for self-driving vehicles.