GenAI Consumes Lots of Energy, But What is Net Impact?

Much has been made of a recent study suggesting ChatGPT operations consume prodigious amounts of electricity, as exemplified by the claim that ChatGPT operations consume 17,000 times more energy than a typical household.  

No question, cloud computing requires remote data centers, and data centers are big consumers of energy. In the United States, data centers now account for about four per cent of electricity consumption, and that figure is expected to climb to six per cent by 2026, according to reporting by The New Yorker. 

But that is not the whole story. Data centers, apps and cloud computing are used to design, manufacture and use all sorts of products that might also decrease energy consumption. Some would argue, for example, that there is a net energy reduction when people use ridesharing instead of driving their personal vehicles. 

Study Title	Location	Key Findings
Life Cycle Energy Consumption of Ride-hailing Services: A Case Study of Taxi and Ride-Hailing Trips in California (2020)	California, USA	- Ridesharing resulted in 11-23% lower energy consumption compared to private vehicles, primarily due to higher vehicle occupancy.
The Energy and Environmental Impacts of Shared Autonomous Vehicles Under Different Pricing Strategies (2023)	N/A (Hypothetical Scenario)	- Shared Autonomous Vehicles (SAVs) with high occupancy rates have the potential for significant energy savings compared to private vehicles.
Future Transportation: The Social, Economic, and Environmental Impacts of Ridesourcing Services: A Literature Review (2022)	N/A (Literature Review)	- Ridesharing can potentially reduce vehicle miles traveled (VMT) compared to private vehicles, leading to lower energy consumption. - However, concerns exist regarding: * Increased empty miles driven by rideshare vehicles searching for passengers. * Potential substitution of public transportation trips with ridesharing, negating some environmental benefits.
Life-Cycle Energy Assessment of Personal Mobility in China (2020)	China	Ridesharing with three passengers can reduce energy consumption.
The Energy and Environmental Impacts of Shared Autonomous Vehicles (2021)	N/A	Shared autonomous vehicles can reduce energy consumption.
Empty Urban Mobility: Exploring the Energy Efficiency of Ridesharing and Microtransit (2019)	Europe	High-occupancy ridesharing reduces energy consumption, compared to use of private vehicles, but we also must account for energy consumed when not transporting passengers.

So far as I can determine, nobody has really tried to model the net energy impact of generative artificial intelligence, data centers or cloud computing, where the energy footprint of GenAI, data centers or cloud computing is compared with the possible net reductions throughout an economy if the app outputs are used to reduce energy consumption in products using cloud computing, data center and GenAI  outputs. 

Study Title	Key Findings
Green Cloud? An Empirical Analysis of Cloud Computing and Energy Efficiency (2020)	Cloud computing adoption improves user-side energy efficiency, particularly after 2006. - SaaS (Software-as-a-Service) contributes most significantly to both electric and non-electric energy savings. IaaS (Infrastructure-as-a-Service) primarily benefits industries with high internal IT hardware usage.
The Internet: Explaining ICT Service Demand in Light of Cloud Computing Technologies (2015)	Cloud computing can lead to increased energy consumption in data centers. Potential for energy savings in other sectors due to: * Reduced need for personal computing devices.  Improved resource utilization and consolidation.
Decarbonizing the Cloud: How Cloud Computing Can Enable a Sustainable Future (McKinsey & Company, 2020)	Cloud adoption powered by renewables can significantly reduce emissions compared to on-premise IT infrastructure. Cloud enables the development of various sustainability solutions (smart grids, remote work).
Cloud Computing: Lowering the Carbon Footprint of Manufacturing SMEs? (2013)	Case studies of manufacturing SMEs shifting to cloud-based solutions.

But some related research suggests ways of looking at net energy footprint. 

Industry	Cloud-based Solution	Potential Fuel Savings	Source
Trucking	Route optimization with real-time traffic data	Up to 10%	DoT
Railroad	Predictive maintenance for locomotives	5-10%	Wabtec
Shipping	Optimized container loading and route planning	5-15%	Massey Ferguson

The point is that “net” impact is what we are after.

Open Interpreter Can be Used for Writing Code

It is not a capability I am likely to need from my own AI use cases, but generating code is among the salient use cases for large language models. 

AI Revenue in the "Picks and Shovels" Phase

Some idea of the ultimate creation of revenue within the artificial intelligence ecosystem, broadly defined to include many of the same participants in the internet ecosystem will shed light on ultimate outcomes. 

Right now, the clearest identifiable revenues are earned by providers of infrastructure, such as graphics processing units or providers of computing-, storage-, models- or inferences-as-a-service. 

In the U.S. internet ecosystem, for example, some $3.3 trillion is estimated to be earned annually by firms in the ecosystem. As always, many of the products in the ecosystem use the internet in some way to support the features and value of their products. 

Not all semiconductors, devices, advertising, platforms or e-commerce and content, for example, are directly and fully attributable to “the internet” in specific. 

In some cases, as with home broadband, it is infrastructure that continues to supply much of the revenue. But even there, revenue earned supporting the use of mobile devices produces the majority of revenue, and internet access is only part of the value proposition. 

Segment	Estimated Annual Revenue (USD Billion)	Source
Semiconductors	200	[1, 2]
Devices	500	[3, 4]
Apps	150	[5, 6]
Advertising	300	[7, 8]
Platforms	500	[9, 10]
Connectivity	250	[11, 12]
E-commerce	1.2 trillion	[13, 14]
Content	200	[15, 16]
Total	3.3 $Trillion

Sources:

[1] Semiconductor Industry Association: https://www.semiconductors.org/

[2] Gartner: https://www.gartner.com/en/newsroom/press-releases/2023-04-26-gartner-forecasts-worldwide-semiconductor-revenue-to-decline-11-percent-in-2023

[3] Consumer Electronics Association: https://www.cta.tech/

[4] IDC: https://www.idc.com/

[5] App Annie: https://go.appannie.com/AppAnnieLite_IWD_Form.html

[6] Sensor Tower: https://sensortower.com/

[7] Interactive Advertising Bureau: https://www.iab.com/

[8] eMarketer: https://www.insiderintelligence.com/topics/category/emarketer

[9] Meta Platforms Investor Relations: https://investor.fb.com/home/default.aspx

[10] Alphabet Investor Relations: http://abc.xyz/investor/earnings

[11] CTIA - The Wireless Association: https://www.ctia.org/

[12] Federal Communications Commission: https://www.fcc.gov/

[13] Digital Commerce 360: https://www.digitalcommerce360.com/

[14] Statista: https://www.statista.com/

[15] National Endowment for the Arts: https://www.arts.gov/

[16] Bureau of Labor Statistics: https://www.bls.gov/

For such reasons, the ultimate winners within the AI ecosystem will likely be more difficult to identify with precision. Specific AI hardware and software will continue to be a source of revenue for some providers within the value chain. 

But most of the AI-related revenue upside will come in more-indirect ways, as AI becomes a feature of many products and services but not a direct and identifiable revenue stream. 

All GPTs Amplify Some Human Capability

All general-purpose technologies amplify some human capability. At least some GPTs seem to amplify multiple capabilities, which makes it hard to predict how the GPT will have impact. And that is true for artificial intelligence as well, assuming it does develop as a GPT.

Consider the many ways the internet created platforms for new ways of doing things; new industries and firms. 

Some of us might note the flattening of hierarchies; value chains with less friction; speed and liquidity of transactions and communications the internet has enabled. Think of all forms of e-commerce as examples. 

Much of that benefit happened because the internet eased the time and cost to obtain information and share it. Think about search as a prime example. 

Likewise, problem solving is easier, to the extent we can tap large volumes of historical data using new tools for data analysis, visualization or simulation. Though the internet was not solely responsible for those developments, it slashed the cost of doing all that. Think about the benefits of cloud computing and capabilities “as a service.”

So price discovery became more transparent; distribution could be made simpler, faster, cheaper; markets could expand on a wider geographic basis, leading to economies of scale or scope. Think of all forms of resource sharing such as Uber or Airbnb. 

Instant communication and connection with anyone, anywhere in the world has enabled new forms of collaboration across borders and an increase in the speed with which ideas can propagate. Instant multimedia messaging and low-cost visual communications are examples. 

To a great extent, the internet has increased everyone’s ability to create and share content, as well. Social media and online media provide examples. 

We might typically think of the value of AI as leveraging thinking rather than muscle power. But past GPTs have reshaped several capabilities. 

Electricity extended our active hours with artificial light, so created, in a sense, “more time” by essentially enhancing “vision.”

Electricity enabled refrigeration, so we got better food preservation and variety, and so enhanced “taste and smell.”

Electricity enabled new medical equipment, leading to better health and longevity

Eventually, electricity led to advanced cognitive abilities as it enabled computers and communication technologies. 

The internal combustion engine increased mobility and transportation speed, so revolutionized agriculture and manufacturing, essentially multiplying muscle power. That aided the growth of global trade and economic growth,

Steam power provided an earlier instance of enhancing muscle power, leading to mass production and mechanized industrial production, as well as steamboats and railroads. Higher living standards were an outcome. 

The internet has enhanced most senses (sight, hearing); “speech” (the ability to communicate) and a variety of cognitive capabilities (collaboration; information access; memory; art)

Artificial Intelligence is likely to reshape cognitive, physical and creative realms, affecting a wide range of human capabilities. That, in turn, might lead to a rather-vast array of outcomes beyond the obvious “who wins and who loses” angles as they apply to employees, firms and industries. 

As we could not predict the rise of search, social media, sharing economics or information technology with an advertising business model when the internet arose, so we cannot predict what big new business models, industries or firms might arise as AI propagates. 

Investors, entrepreneurs and incumbent firm executives will try, though. “Faster, cheaper, better” might be an easy way to describe AI impact on existing processes and activities. But it is the unexpected and big new developments that people will seek, fear and act upon, even when we aren’t sure what those might be.

Can AI Fix "Tags?"

If you are a person who writes hundreds of blog posts every year, you have encountered the tedium of tagging or keywording or otherwise classifying content. The problem is similar to the issue with structured databases, which requires predefined fields and relationships. 

One has to choose a limited number of tags, based on assumptions about what other users will be searching for. But searched-for tags will shift over time, especially when new relationships or connections are investigated. 

That chore of pre-defining what tags will “always” make sense is so imprecise and unpredictable that I long ago gave up using them. Yes, that might make the content harder to “find,” but it saves so much manual work that the tradeoff I deem reasonable. 

In principle, artificial intelligence should solve some of the problem, not only by automatically classifying content, but by creating something like the value of a relational database compared to older, more-static models where adding new tables or relationships without fear of disturbing what already exists. 

Older databases, for example, were hierarchical, where data is organized in a parent-child structure, making complex queries challenging. Flat-file databases also limited querying capabilities. 

In principle, AI should provide similar benefits for tagging functions. There should be less manual labor to classify information in a rigid way, where relationships are not always pre-defined. That should, in principle, allow future searches to use more-dynamic queries that go beyond the fixed structure. 

Equally important, AI should enable easier discovery of new connections beyond the predefined keywords or database schemas. And where assignment of tags and keywords can be subjective and ambiguous, AI should allow us to overcome such ambiguity or subjectivity. 

Tags and labels are often applied based on individual interpretations, leading to inconsistencies and ambiguity, especially across large datasets. AI should help overcome such limitations.

Limited scope and granularity could be less an issue after we apply AI. And, obviously, less manual effort will have to be expended for content classification. And AI-enhanced translation should mean all content can be accessed, irrespective of original language. 

Of course, many will note that AI still lacks the fuzzy, subtle and contextual nuances that make expert human tagging useful. But it should be better than manual tags.

And what would ultimately be better is automated classification that updates over time as "key words" change, and the original tags lose some relevance, while possibly becoming more relevant in new contexts.

AI Phone Replaces Apps?

Almost by definition, artificial intelligence devices beyond the smartphone (pins, pendants, other form factors) will need to operate “beyond apps” as the primary interface will be the spoken word and screens might or might not be used.

The interface will take the form of commands or questions that generate responses that are not confined or based on use any one installed app. 

Mobile World Congress, for example, T-Mobile showed an AI phone concept created with Qualcomm Technologies and Brain.ai featuring an AI assistant that replaces apps used on the smartphone.

How Big is the Telco API Opportunity?

According to Ericsson, connectivity service providers have the opportunity to capitalize on the expansion of the communication platform as a service market--essentially voice or messaging enabling any other apps--by using application programming interfaces.

In part, that is an effort to reclaim some revenues being earned by other third-party providers of such capabilities, such as Twilio. Twilio makes money by providing a suite of APIs  that allow developers to integrate various communication channels (voice, SMS, video, email) into their applications.

The revenue model then typically is based on usage fees: Developers are charged based on their usage of these APIs, including per-minute voice calls, per-message sent, for example. 

The opportunity for connectivity service providers might be as high as $94 billion in annual revenue by about 2030, a target large enough to be interesting for mobile operators. 

Platform	Study Name	Date of Forecast	Publishing Venue	Revenue Forecast (Billions USD)
Twilio	Twilio Investor Relations	February 2024	Company Website	$5.84 in 2023, expected to reach $14.3 by 2026
Vonage	Gartner Magic Quadrant for UCaaS, Worldwide	February 2024	Gartner Report	$2.23 in 2023, projected to grow at 8.8% CAGR to reach $3.48 by 2028
Plivo	Plivo Blog: The State of Communications APIs in 2024	February 2024	Company Blog	$400 million in 2023, expected to reach $1.2 billion by 2027
Sinch	Sinch Annual Report 2023	February 2024	Company Website	€510 million in 2023, anticipated to reach €1.2 billion by 2026
Bandwidth	Market Research Future: Unified Communications as a Service Market 2023-2030	February 2024	Market Research Report	$26.71 billion in 2023, predicted to grow at 22.7% CAGR to reach $93.62 billion by 2030 (This includes all UCaaS providers, not just Bandwidth)

Though a controversial move at the time, the acquisition aimed to leverage Vonage's CPaaS offering based on Application Programming Interfaces to expose network capabilities. 

Ericsson aimed to leverage APIs originally for enterprise use cases, but now also seemingly is embraced by some mobile operators as a way of creating new revenue streams based on voice- or text- or “other network features” enhancing apps. 

Vonage's large developer community also was viewed as a value. 

The larger point is that mobile operators might now view CPaaS as a reasonably attractive revenue growth opportunity, even if all that happens is that market share is taken from other providers.