Generative AI is Mostly Enhancement, Not Disruption, in Most Instances

It is a reasonable enough assumption that an AI-enhanced Alexa will represent an incremental enhancement rather than a disruptive change, at least at first.  And perhaps the same might be said for AI-enhanced search, social media, advertising, e-commerce customer service or virtually any other process that underlines a business model. 

In most cases, generative artificial intelligence will be used to upgrade or enhance existing apps and use cases. 

Perhaps of more interest are ways GenAI might be used to support entirely new use cases, such as AI-driven or automated art generators, music composers, and story writers. Of course, some might argue that AI-generated art is an extension of existing computer-generated graphics functions.

Generative AI might also be used to accelerate and automate drug discovery by generating new molecular structures and predicting their properties.

GenAI also can be used to generate realistic gaming environments, characters, and storylines.

So far, there is not widespread agreement on whether GenAI can be a platform for entirely-new use cases with new business models. 

A related question is whether third-party models or in-house models will drive such developments. 

Many use cases already are being built on third-party models. Siri using a platform provided by Anthropic’s Claude provides an example. The use of third-party platforms rather than in-house models is a choice most firms likely will be making, and for reasons similar to their use of any important technology. 

Core firm competence almost never lies in the area of operating system, computing appliance or platform development. Also, time-to-market concerns plus performance make “build your own” approaches either too time-consuming or too expensive or both. 

Most firms have no interest in building their own chips, operating systems, core apps, computing hardware, networks or AI models. But lots of firms might have interest in customizing existing third party models for a particular industry business process. 

Firm	Model Licensed	Use Case or Application
Apple	OpenAI's GPT-3	Siri Enhancements, Text Analysis
Amazon	Anthropic's Claude	Customer Service, Product Recommendations
Microsoft	OpenAI's GPT-3	Bing Chat, Office Suite Enhancements
Meta	Llama 2 (open-source)	Research, Product Development
Salesforce	OpenAI's GPT-3	Customer Relationship Management, Salesforce Einstein
Snapchat	OpenAI's GPT-3	AI-powered Lenses, Chatbot Features
Spotify	OpenAI's GPT-3	Music Recommendations, Podcast Summaries

Mobile Generative AI Will Be a Huge Driver of Usage

Eventually, the leading generative artificial intelligence apps used by consumers will winnow down from hundreds to a few, though many platforms will likely find niche uses in some industries, market segments, job functions and use cases.

One important difference could arise in the mobile domain, compared to larger-screen use cases, and the reason is simply the huge amount of interactive app usage that now happens on mobile devices, compared to all other screens. 

User Type	Mobile Devices	PCs
Consumers	5-6 hours/day	2-3 hours/day
Business Professionals	3-4 hours/day	5-6 hours/day

Perhaps more important is the amount of data consumed on mobile platforms. While it might be difficult to directly correlate “value” with “data volume,” data consumption is connected with usage volume. Generally speaking, the volume of consumer data used on mobile devices is twice as much as on PCs, for example. 

User Type	Mobile Devices	PCs
Consumers	10-20 GB/month	5-10 GB/month
Business Professionals	20-30 GB/month	10-20 GB/month

That usage is then correlated with the volume of advertising spending on mobile and PC platforms. 

Among the big shifts in U.S. advertising spending in the internet era is not simply the growth of share taken by digital media, but also the share taken by mobile venues, which already are as much as 65 percent of all digital media advertising. 

Venue	Spending  (Billions USD)	Market Share (%)
Digital	233.4	56.7
Television	73.7	17.9
Radio	21.9	5.3
Out-of-Home	19.5	4.7
Print	15.6	3.8
Other	12.7	3.1

source: Statista, Seeking Alpha 

That suggests a huge opportunity for generative AI  use cases on mobiles, as well, assuming that GenAI winds up being a core functionality of most highly-used consumer-facing apps. 

source: Andreessen Horowitz 

source: Andreessen Horowitz 

And to the extent that the costs of GenAI usage for app and experience providers matters, open source models should, over time, increase their share of the market. Already, in mid-2024, enterprise user /.,mnbvcz+net promoter scores for open source GenAI models have rapidly approached those of proprietary models. 

source: Andreessen Horowitz 

As always for new markets, early market share often is not predictive of developed or mature market leadership. Eventual leaders often are not among the early leaders. Also, definitions of “active” users might vary. Some might include users who have accessed a model only once; others will have varying levels of persistent usage that are minimums for the purpose of defining active users. 

Model	Estimated Active Users
OpenAI (GPT-4, GPT-3.5)	100+ million
Google (LaMDA, PaLM, Gemini))	50+ million
Meta (LLaMA)	30+ million
Microsoft (Azure OpenAI Service)	20+ million
Stability AI (Stable Diffusion)	10+ million
Midjourney	10+ million

The term "active user" can include:

• Frequency of Use: How often a user interacts with the AI model. This could be measured by the number of prompts or requests submitted within a specific timeframe (e.g., daily, weekly, monthly).

• Duration of Use: The amount of time a user spends interacting with the AI model during a session or over a period.

• Type of Interaction: The nature of the user's interactions, such as text prompts, image generation requests, or code completion.

• Conversion Rate: The percentage of users who take a desired action, such as creating an account, subscribing to a premium service, or sharing content.

• Engagement Metrics: Measures of user engagement, such as click-through rates, time spent on the platform, and social sharing.

For OpenAI, “active users” are typically defined as those who have interacted with the model within a specific timeframe, such as the past month or year.

Google's definition of "active user" considers frequency of use and engagement metrics. Meta's definition of "active user" might be similar to its definitions for other products, focusing on user engagement and interaction.

Verizon Flips Assets: Selling then Buying Frontier Communications

Asset flipping in any business is not unheard of, but Verizon’s history with Frontier still seems instructive. In 2010, for example, Frontier Communications purchased rural operations in 27 states from Verizon, including more than seven million local access lines and 4.8 million customer lines. 

Those assets were located in Arizona, California, Idaho, Illinois, Indiana, Michigan, Nevada, North Carolina, Ohio, Oregon, South Carolina, Washington, Wisconsin and West Virginia, shown in the map below as brown areas. 

Then in 2015, Verizon sold additional assets in three states (California, Texas, Florida) to Frontier. Those assets included 3.7 million voice connections; 2.2 million broadband internet access customers, including about 1.6 million fiber optic access accounts and approximately 1.2 million video entertainment customers.

source: Verizon, Tampa Bay Business Journal 

Now Verizon is buying back the bulk of those assets. There are a couple of notable angles. First, Verizon back in the first decade of the 21st century was raising cash and shedding rural assets that did not fit well with its FiOS fiber-to-home strategy. In the intervening years, Frontier has rebuilt millions of those lines with FTTH platforms.

Also, with fixed network growth stagnant, acquiring Frontier now provides a way to boost Verizon’s own revenue growth. The acquisition means Verizon’s FTTH  connections will jump from approximately 7.4 million to 9.6 million, a gain of about 23 percent in one fell swoop. And since home broadband is the primary revenue growth driver for fixed networks these days, that matters. 

source: Verizon 

There are other takeaways. As in the mobile communications business, where Verizon and AT&T, for example, had been focusing on urban footprints and customers, market saturation has forced both firms to plumb rural areas and customers as well as the mobile virtual network operator business and prepaid accounts, where the main focus had been postpaid branded accounts, market saturation has forced the major providers to search in new areas for growth. 

As a byproduct, Verizon might, in some cases, be able to leverage its new fixed network assets to support its mobile network as well (fiber backhaul, for example). 

It is possible there are other strategic considerations as well. T-Mobile, which started out with zero share of the fixed network home broadband market, now is growing based on its use of fixed wireless services provided by its mobile platform.

But T-Mobile is making its first steps towards adding some amount of fixed network access provided by cabled networks as well. For example, T-Mobile has partnered with EQT, a global investment firm, to acquire Lumos, a fiber-to-the-home platform.

T-Mobile also formed a joint venture with KKR, another global investment firm, to acquire Metronet, a leading fiber-to-the-home provider. That acquisition also will expand T-Mobile’s fixed network home broadband market share.

And while it has seemed unlikely that T-Mobile would contemplate moves such as acquiring Frontier Communications or other firms such as Brightspeed itself, that outcome--at least regarding Frontier--is closed. 

On the other hand, the pressure to grow footprint to grow market share remains intact. Brightspeed does appear to have substantial overlap with Verizon’s new fixed network footprint, but duplicated assets might be sold. 

And Verizon appears to face little danger of antitrust action were it to acquire additional fixed network assets, given its modest coverage of U.S. homes. By some estimates, prior to the Frontier acquisition,   

Verizon homes might have numbered less than 25 million, possibly as low as 20 million. 

That is far fewer than top Verizon competitors might claim. 

Comcast has (can actually sell service to ) about 57 million homes passed. AT&T’s fixed network represents perhaps 62 million U.S. homes passed. 

Charter already passes more than 32 million locations, including homes and businesses. 

CenturyLink never reports its homes passed figures, but likely has 20-million or so consumer locations it can market services to.

The point is that additional Verizon acquisitions of fixed network assets, to reach more U.S. homes, might not pose antitrust issues. The Frontier acquisition adds between five million to 10 million potential new fixed network locations (not all upgraded for FTTH, yet, and including business locations). That potentially increases Verizon’s “locations passed” footprint by as much as a third. 

Using Verizon’s recent assertion that, after the Frontier acquisition, Verizon will reach 25 million homes, Verizon would still have some ways to go before it passes as many homes as AT&T, Comcast or Charter, its larger fixed network competitors. 

Frontier is said to have a network reaching 15 million locations, including homes and businesses. A reasonable guess is that at least 10 million of those locations are homes. 

Most of those locations are arguably not good candidates for FTTH investment, which is why firms such as Verizon and Lumen sold off rural footprints in the past. 

If Verizon’s “homes passed” footprint, after the acquisition, is only 25 million, there remains room to add more homes by acquisition.

Brightspeed’s network seems to pass about 6.5 million locations. Most are homes, but not all. Assuming 90 percent are residential, that implies less than six million locations are homes. So even adding Brightspeed assets would only bring Verizon up to perhaps 31 million or so homes, still far less than reached by AT&T, Comcast and Charter. 

The point is that the strategy of selling off rural assets and re-acquiring them later, once a critical mass of FTTH passings and accounts have been created, seems a logical strategy. Verizon’s cost to acquire the Frontier footprint (not customers, but network passings) is north of $1,000 per location, and possibly in the $1500 per passing range. 

Many observers expect that the former Frontier FTTH passings will double within a couple of  years. At current take rates, that also implies a potential additional two million or more FTTH accounts being added. 

Asset flipping remains part of the connectivity business. But it is rare to see a seller reacquire its sold assets.

Comparing GenAI to PC Impact on User Interface

Generative artificial intelligence might be likened to several elements of personal computing. Just as personal computers revolutionized how individuals interacted with technology, generative AI is poised to fundamentally change the consumer relationship with machines. Machine learning, in contrast, remains largely an enterprise use of AI. 

Both personal computers (more aptly, the applications and software consumers can run on PCs) and generative AI models are accessible to a wide range of users, for “daily life” tasks,  regardless of their technical expertise. 

User-friendly interfaces (graphical user interfaces, for example; “what you see is what you get” displays); multimedia content; portability and mobility and a focus on content (social media; driving directions; answers; price discovery; product or service availability; entertainment; ordering and learning) made PCs the first consumer-focused computing experiences. 

Generative AI, with its focus on content and conversational interface, likewise makes it useful for consumers in applications including entertainment; learning; answers; ordering; price discovery and availability; o business and research.

Similarly, generative AI is being applied to a wide array of consumer tasks, including content creation, customer service, and scientific discovery.

Also, the GenAI adoption context is quite different from that of the first PCs. Unlike PCs, which required something of a paradigm shift in how people interacted with technology, generative AI often enhances tools people are already using. So where PCs were largely disruptive or revolutionary, GenAI is mostly incremental and evolutionary. 

Early PCs had no keyboards, monitors or convenient storage. They required extensive hacker skills; investment in new hardware and did not immediately solve any common consumer problems. GenAI is almost the polar opposite, enhancing and improving lots of experiences consumers already are using, using existing hardware and platforms.  

Natural language is similar to the graphical user interface in terms of enabling an easy human interaction with the machine and its software. 

So writing assistants improve or enhance word processors; enhance photography; improve search and e-commerce or social media personalization. AI assistants extend capabilities of existing voice interfaces and appliances. 

Where personal computing originally had high barriers to entry (which became far lower with graphical user interfaces; point and click), genAI has low barriers to entry, using conversational language interactions on existing software and applications and existing hardware and platforms.