How to Make Money with AI – Part 1: ROAI

(Originally published July 4, 2023 on LinkedIn)

Today is Independence Day. The 4th of July (2023).

I am publishing this article today in honor of the United States of America. USA indisputably stands as the greatest country on the planet. USA is the birthplace of free market capitalism and countless technology innovations that can potentially improve the lives of every person on the planet. Pioneering tech entrepreneurs and Founders, fueled by relentless grit and determination, have decisively solidified the USA’s leadership in the global arena, delivering on the promise of new tech breakthroughs and demonstrating the unbridled power of free society. This article is at the intersection of tech innovation and free market capitalism. How to make money with AI – it’s the ultimate question.

Intro – Historical Science & Technology

The most recent tech transformation in 2023 is AI (Artificial Intelligence). You may have heard about it already – it’s in all the headlines these days ;-)

I believe history books will attribute November 30, 2022 (public release of ChatGPT by OpenAI) as the official birth of the AI era. Previous breakthroughs in science and technology ultimately affecting every person on the planet before AI include:

  • Humans Controlling Fire – c. 500,000 BC
  • Printing Press – 1440
  • Electricity – 1879
  • Personal Computers 1980s
  • Internet (Netscape IPO) – 1995
  • iPhone (mobile smartphones) – 2007

All these pre-AI technological breakthroughs resulted in massive opportunities for making money. However, in the opinion of this veteran Internet entrepreneur, AI now represents the single most-powerful force in human history for making money. Almost everyone in tech is currently obsessing over the best ways to make money in this new era of AI. The ultimate question is – as it has always been – how to make money using this new technology? The ultimate answer is unfortunately not 42.

Internet + Google = Make Money

As an old Internet tech entrepreneur that started in the mid 1990’s, I believe that the best answer for how to make money using AI is by analyzing how Google made money in the new era of the Internet, beginning in 1995 with the Netscape IPO – the “birth” of the consumer Internet. Google is objectively the single most successful Internet business ever in terms of growth, value creation, and value capture – i.e., making money.

Google’s consistent revenue growth, profitability, and cash flows are unmatched by any other Internet tech business, and these achievements are largely a result of its business model and strategies. Google was founded in 1998 as the ~12th Internet search engine. When Eric Schmidt joined Google as CEO in 2001, the company was not fully certain which revenue model would yield the highest returns, but ultimately bet on an Internet Advertising business model over Internet Search hardware/services sales into the enterprise (who among the readers of this article remembers the Google Search Console box sold to enterprises and data centers? ;-)

Google’s ad platform turned the business world on its head by helping Google achieve an unprecedented $1 billion in revenue in record time (Google first reported annual revenue of over $1 Billion in 2003), simply by selling Internet ads to businesses. The bet on Internet Ads was one of the most impactful decisions of early Google, as represented by this table below of total annual revenue of Internet business models in 2020 (i.e., business models that could not exist without the Internet, thus excluding e-commerce, travel bookings, etc. since those models could conceivably operate without an Internet):

2020 Revenue by Internet Business Model Category

It is worth also noting that Google Cloud represents a meaningful portion of the Cloud Computing Services revenue as of 2020, but the Google Cloud offerings were launched years after the Internet Advertising business was firmly established and blissfully cash-flowing. For the more “mission-driven” and “societal-impact” readers of this article, it is also worth noting the blissful cash-flows from Google’s Internet Advertising business has funded countless mission-driven societal-impact initiatives including self-driving vehicles, autonomous drones, Internet accessibility in remote areas, life sciences and health technologies, biological research on lifespan, etc. Personally, I am a huge fan of this “first, get the money” model used by Google, and only then investing in more noble impact initiatives after that.

“Selling Growth” – Startup Growth via Google Ads

Rather than attempting to compete directly with Google, Internet founders/entrepreneurs saw a different opportunity – BUY from Google rather than compete against them. Startups realized the power of buying ads from Google to propel the growth of their new Internet businesses. The model effectively became a high-yielding arbitrage play: paying for cost-per-click (CPC) ads from Google and capitalizing on the high value-per-arrival (VPA) on their landing pages.

This strategy was especially effective in high customer lifetime value (LTV) verticals like finance, travel, insurance, home services, and online for-profit education. Google advertisers ingeniously turned the act of buying traffic into a virtue, emphasizing their data-driven growth and efficient customer acquisition strategies. These industries and verticals represented Google’s highest ad revenue markets for this very reason.

Startup businesses recognized the optics of their dependence on Google ads for their growth as a down-side to their enterprise value, but controllable/scalable growth and solid unit economics represented a solid/calculated tradeoff for the dependency on Google traffic. A well-balanced growth plan would ultimately require startups to gradually build their own branded/organic traffic to reduce reliance on paid traffic channels like Google, but that was often deferred to later stages of the business. Thus, the history of startups growing via Google paid search turned out to be a strategy virtue not a flaw. A planned feature, not a bug.

Internet Advertising is intrinsically a B2B sale. Businesses that spend money on Internet Advertising sell their goods and services to either consumers or to other businesses. A disproportionate amount of Internet Advertising sales has historically been to businesses that sell to consumers (B2C), which is congruent with the US economy being 70%+ a result of consumer spending.

Additionally, it is estimated that 75%+ of Google Internet Advertising sales are for net new “Customer Acquisition” efforts, including e-commerce, phone sales (i.e., lead generation), ad arbitrages, app installs, and subscriptions. These advertisers are considered “Performance Marketers” because they are solely focused on the profitability of their ad spend (i.e., ROAS – Return-on-Ad-Spend). The remaining ~25% of Google ad sales are considered non-Customer Acquisition spend, primarily for branding and data collection without a clear ROAS measurement associated with the spend.

For all types of businesses advertising for the purpose of Customer Acquisition – especially for businesses selling to consumers – the ultimate goal and desire of the business is revenue growth. It’s the holy grail. Google’s success is a direct result of the growth that their Internet ads enabled for their advertiser/business customers. By demonstrating that using Google ads results in a measurable increase in advertiser’s own sales and revenue, they tapped into the primary desire of all their business customers. Google was essentially selling “Growth,” which is something all businesses crave. By doing so, Google grew faster than any other business in history, while making unprecedented amounts of money in the process.

Analysis & Framework of Google Unprecedented Success in Making Money

With the benefit of hindsight, we can now analyze all of Google’s decisions, strategies, and business models to infer repeatable methods of how to make money on the Internet. I believe that this framework can then be applied to methods of how to make money using AI – the single greatest money-making technology breakthrough in human history – as a template:

  1. Consumer-facing = Maximizing the Value of the Internet – For Google, every single new Internet user represented incremental revenue opportunities via Internet Advertising. Using Internet Ads, Google and other businesses could conceivably communicate (and potentially interact) with every single Internet user on the planet. Virtually unlimited distribution. Thus, Internet Advertising became the “killer money-making app” on the Internet – free for consumers but hundreds of billions in annual revenue from advertisers. Technologies that are not consumer-facing confront inherently lower ceilings of their money-making potential, since their total customers/users and/or audience will be a small subset of the “all Internet users on the planet” that Google can claim.
  2. Sell B2B2C – The B2B2C (Business-to-Business-to-Consumer) model creates value that is greater than the sum of separate B2B (Business-to-Business) and B2C (Business-to-Consumer) models. It is considered by many to be the most powerful, profitable, and fastest-growing of all Internet business models. A B2B2C model like Google uses allows for the potential monetization of billions of individual Internet users (consumers) without needing billions of individual business relationships/accounts. Google can benefit from the monetization of millions of consumers vis-a-vis a single advertiser customer account, which drastically slashes the cost and effort of servicing the billions of dollars in revenue.
  3. Sell “Growth” – 75%+ of Google’s Internet Advertising sales are a result of net-new Customer Acquisition efforts by their advertisers. Thus, Google’s primary/core value proposition to businesses is to MAKE money, not SAVE money. This is playing offense with new technology, not defense. Conversely, most B2B enterprise software sales value propositions are to SAVE money on live human labor costs. B2C businesses that buy advertising have a clear preference for growth. Cost savings simply don’t generate the same level of excitement as does growth. Revenue growth, however, has a much higher value-creation ceiling (virtually infinite) via increased sales. Revenue growth also offers the potential for exponential gains. For example, a hot new product or a successful customer acquisition campaign can lead to a rapid increase in customers and sales, leading to a multiplier effect on revenue. Also, from a human psychology perspective, growth is exciting. Growth is FUN :-) Growth clearly indicates progress, innovation, market dominance, and increased value, power, and influence. Growth is an indicator of success that everyone – from Founders and investors to employees and customers – can get behind. The best thing to sell to businesses is “Growth.” QED.
  4. 100% Software Platform – Google’s incredible profitability and cash-flows are a result of their software platform business model where zero live humans are required for daily revenue production end-to-end. Value is created by facilitating exchanges between advertisers (producers) and Internet users (consumers). The systems are automated, and they leverage algorithms and machine learning to serve users and advertisers effectively at the relatively low cost of moving zeroes and ones around on the Internet at the speed of light. Electrons.
  5. Command Value-based Pricing – Google uses an auction-based pricing model. This system creates “scarcity” because advertisers are essentially bidding for a finite number of Internet users (consumers). Scarcity comes into play because there is a finite number of users available. As more advertisers bid, their respective “price-per” costs tend to increase, reflecting the increased competition for consumers relative to the perceived scarcity of the consumers. Thus, pricing is determined from the top-down, based on the Expected Value (EV) of the advertisers, rather than determined by the cost to Google to deliver the ad with an arbitrary profit margin applied on top (i.e., bottom-up). This value-based pricing is a key method for Google to capture the maximum amount of value that they create via profits derived from the efficiency of the marketplace resulting from the advertisers’ bids. In stark contrast, airlines use a cost-up pricing model which prevents them from capturing the true value of their passengers’ flights, and prevents the airlines from making money (profits).
  6. ROAS = Supreme Money-making Metric – ROAS (Return-on-Ad-Spend) represents a clear supreme metric for measuring the financial performance of money spent on Google Advertising. The “Ad Spend” part of ROAS is absolute. It’s an objective and fixed amount at any point in time. “Return” on the ad spend can be calculated any way the advertiser believes is optimal to convey the financial performance of the ad spend, as long as it is quantifiable. The “Return” value in the numerator could be revenue (most common), profit, or some other quantifiable metric of “Return” used by the advertiser. Please note the “Ad Spend” in this context is customarily considered OpEx and/or COGS on the advertiser’s P&L, and seldom capitalized on the balance sheet. This will be an important distinction later in this article when we consider the ROAI as the supreme metric for measuring financial performance of AI.

Applying Google Money-Making Success Framework to Making-Money with AI

By acknowledging Google as the single greatest money-making business in history as a result of the transformative new technology of the Internet, and by analyzing Google’s framework of early decisions and strategies, we can infer and inform a similar business that has the best opportunity to become the new greatest money-making business in history – but this time, using the transformative new technology of AI. I believe that the greatest money-making AI business startup in 2023 should reflect the early Google framework:

  • Consumer-facing (70%+ of the US economy)
  • Sell B2B2C (monetize millions/billions of consumers via relatively small number of customer accounts)
  • Sell “Growth” (75%+ of Google Ad spend is for Customer Acquisition efforts)
  • 100% Software Platform (no humans required for daily revenue production)
  • Command Value-based Pricing (capture maximum amount of value created)
  • ROAS = Supreme Money-making Metric (the key to recurring revenue – ongoing “proof” of financial performance)

Symphony42 uses ALL these pages from the Google playbook, but leveraging AI as the transformational tech in addition to consumer Internet traffic that Google leveraged. In this next section, however, I will focus specifically on the last item on the list above. ROAS is the supreme money-making metric and will juxtapose it with ROAI, the NEW supreme money-making metric in the era of AI. I will also juxtapose how Google advertisers leveraged the capital-efficient model of buying traffic for their growth instead of the capital-intensive approach of building organic/proprietary traffic for their growth, which would effectively mean COMPETING with Google instead of being Google’s customer :-S

ROAI is the new ROAS

Return-on-Ad-Spend (ROAS) emerged as the supreme metric for Google advertisers, serving as the justification for their Billions of dollars in ad expenditures. ROAS — i.e., the ratio of the value derived from ad spend relative to the cost of the ads — has proven essential in enabling Internet businesses to grow fast and profitably.

Just as ROAS became the supreme metric for Google ads, Return-on-AI (ROAI) is poised to become the same in the realm of AI startups, particularly in this new era of AI-driven growth and spending. As investments in AI technology continue to escalate, the establishment of effective metrics to measure the return on AI becomes crucial. The definitive metric for AI return on investment is ROAI.

The development of in-house AI solutions entails significant capital expenditures, including costs related to hardware, software, and labor. However, companies that use AI platforms such as ChatGPT from OpenAI can shift these upfront costs towards de minimis operating expenses (OpEx) and away from expensive CapEx. This cost dynamic mirrors the “Ad Spend” in ROAS, thereby positioning ROAI as the critical cost unit for evaluating the returns from AI usage.

ROAI is primed to become the preeminent metric for monetization, serving as the key validation for recurring revenue and providing ongoing proof of financial performance. Much like ROAS in the world of Google ads, ROAI is set to become the supreme money-making metric in the domain of AI.

AI Startups – Rent vs. Build

I propose a thought experiment paralleling the growth of Google’s influence on tech startups and the AI startup landscape of 2023…

During the early 2000s, instead of laboriously cultivating organic traffic, many startups chose the more capital-efficient route of buying ads from Google. Building systems for equally predictable growth but from organic traffic instead of traffic from Google ads is a lengthy and capital-intensive process. However, “renting” consumer attention via Google ads represents instant and predictable growth, financed via OpEx/COGS instead of CapEx in a more capital efficient manner.

A similar phenomenon is observable in the AI world today. While some AI startups in 2023 endeavor to develop their own internal proprietary AI models – an expensive and resource-heavy process akin to generating organic traffic – other startups opt for the lean approach. These lean AI startups build their innovative new tech products on industry-leading AI platforms such as the ChatGPT-4 API, a move analogous to purchasing ads from Google for growth traffic. It’s a “Rent” (i.e., license) rather than “Build” (i.e., develop) strategy sometimes criticized for fostering dependence on another business’s platform (e.g., the early Zynga dependence on Facebook), yet its favorable financial risk vs. rewards advantages warrants all startups and VCs consideration in developing appropriately balanced risk/reward strategies.

To illustrate this concept, many successful Internet startups in the early 2000s such as LendingTree, Progressive, Esurance, TripAdvisor, Priceline/, Expedia, and University of Phoenix all became leading examples of this capital-efficient model by buying Google ads to fuel their growth. This approach enabled them to focus their resources on developing and delivering unique value and refining their products instead of obsessing over the herculean challenge of growth via organic traffic. As their businesses expanded, they subsequently invested in their own brand equity and proprietary system to drive organic traffic and reduce their growth dependence on Google. Similarly, many AI startups in 2023 are choosing to start with other AI platforms but will maintain the optionality of building their own proprietary AI models once they’ve achieved certain financial size/scale milestones.

Both approaches can present intriguing prospects, contingent on the startups’ unique offerings and context. While developing an internal proprietary AI model might be a wise choice for a startup with game-changing AI/ML breakthroughs, most may find “Renting” existing APIs, like GPT-4, a quicker and far more capital-efficient route to market. The critical element lies in devising a clear strategy to manage any platform dependence/concentration and guarantee the delivery of a unique value proposition to customers, amidst the evolving AI landscape. In 2023, VCs are increasingly skeptical of “thin product layers sitting on top of ChatGPT.” Also critical to value creation and value capture is the cultivation of valuable/unique proprietary training data – the REAL arms-race in AI – that can be leveraged in the training of all AI models, both internal/proprietary and external.

Also, some AI startups like Symphony42 mitigate the dependency risk by implementing a sophisticated Champion/Challenger methodology, where multiple AI platforms are integrated and used in production simultaneously. This allows startups to diversify their AI dependencies, while identifying the best performers in the market at the lowest cost, and while also always maintaining future optionality for building proprietary AI models after a certain size/scale of the business has been achieved.


As a serial tech entrepreneur, I’m biased, but I genuinely believe that the best way to make money with AI in 2023 is by launching a new AI startup business. My AI startup Symphony42 uses everything analyzed and described in this article, but by leveraging AI as the current transformational tech in addition to consumer Internet traffic that Google leveraged. Additionally, Symphony42 is leveraging the capital-efficient growth strategies of the fastest-growing Google advertisers – rather than building internal proprietary AI systems, which is inherently capital-intensive and occupies a different place on a risk/reward spectrum. The table below details the attributes and strategies of Google, Google Advertisers, and Symphony42.

Business Strategy Attributes of Google, Google Advertisers, and Symphony42

Please provide feedback in the comments section, or contact me directly at :-)

Watch and/or Listen to this article here: