How to Cut the Hype and Profit from AI

For all its promise to transform industries and change the way we live and work, there is still debate in technology and business circles over whether the financial impact of AI lives up to its ambitious claims . As is the case with most technological innovations, the debate is polarized.

On the one hand, business leaders are certain that we are in an AI bubble, much like the Internet bubble of the 1990s and the subsequent stock market crash in 2000. Robin Li, CEO of Baidu, echoed this sentiment at the Harvard Business Review’s Future of Business conference.

“I think like many other technology waves, bubbles are sort of inevitable once you get past the initial excitement stage. People would be disappointed if the technology did not live up to the high expectations generated by the initial enthusiasm.

Li predicted that once this hype passes, only about 1% of companies will survive, ultimately achieving significant growth and creating substantial, lasting value.

On the other hand, venture capitalists like Jon Medved, CEO of OurCrowd — which manages $2.3 billion in assets and more than 460 portfolio companies — see real financial promise in AI. Medved says that AI, like the Internet era, may indeed be experiencing inflated valuations, but the underlying business potential is real.

“Yes, valuations in the dot-com era were ‘bullish’, but the business opportunity was real and great companies like Amazon, Google and many others were created because of it,” Mr. ‘Medved said.

This contrast sets the stage for examining the true business impact of AI today.

State of the AI ​​market

THE The AI ​​market in 2024 is valued with an estimated revenue of $214 billion and is expected to continue growing at a double-digit rate over the coming years, with a projection of $1,339 billion by 2030. These estimates and projections are supported through tangible measures such as increasing corporate budgets for AI tools, growing investments in AI startups, and the growing number of AI-based applications integrated into almost every business function of everyone areas known to man.

In 2023 alone, private AI startups have raised nearly $50 billion in funding and have already taken a large chunk of seed funding for this year, according to Crunchbase. This “AI gold rush,” as ForbesPhoebe Liu called it quits earlier this year, making many tech giants even richer and pushing “more than a dozen new AI billionaires into the world.” Forbes“List of the world’s billionaires.”

But even if the funding is great, how exactly do AI companies make money today?

AI ROI Factors

When it comes to ROI for AI, hardware and infrastructure stand out. Companies like Nvidia have capitalized on demand for advanced computing processors and devices needed to support AI algorithms, particularly graphics processing units, which sell for up to $40,000 per unit, according to Nvidia CEO Jensen Huang. These chips power some of today’s biggest innovations in AI, leading to a possible $5 trillion valuation of the hardware company.

There are two prominent examples of how this translates into revenue for the end user. One offers AI as a Serviceallowing businesses to access AI capabilities such as image recognition, predictive analytics, language processing and more through application programming interfaces without building proprietary infrastructure from zero.

A good example of this is OpenAI’s AIaaS offering, available through the Microsoft Azure OpenAI service and allowing users to access OpenAI’s language models including GPT-4, GPT-3-5-Turbo and others. Using the provided APIs, developers can deploy the GPT-4 model on data and adapt it to various tasks. This revenue source operates on a pay-as-you-go model, which only charges for the resources used.

OpenAI’s most advanced model, GPT-4o, currently costs between $2.50 for 1 million input tokens and $10 for 1 million output tokens – tokens being chunks of words used for processing natural language. With more than a million third-party developers using the technology to power their own services, the San Francisco startup hopes to earn $1 billion from this revenue stream, according to a report. New York Times‘ review of recent OpenAI financial documents.

Also consider Google Cloud AI, another suite of AIaaS products and services that businesses can deploy for specific needs. Some of these solutions include pre-trained machine learning models for image processing, pre-configured discussion platforms using retrieval augmented generation (more widely known as RAG), and a pre-configured solution for extracting text and summarize large documents. These offerings, among others, contributed significantly to the 35% increase in cloud revenue announced by Alphabet for the third quarter of 2024.

Beyond AIaaS, another revenue driver is AI-enabled applications, that is, software that integrates AI for specific use cases. These applications combine AI models with custom industrial data or language models to provide solutions tailored to niche needs. It is under this umbrella that the famous ChatGPT falls.

OpenAI reached $300 million in monthly revenue last August and this came primarily from ChatGPT. For this chatbot, OpenAI offers a flat-rate subscription model for its Plus, Team, and Enterprise plans, starting at $20. But only about 10 million of its 350 million monthly users (a little over 2%) subscribe to the service.

Still, the company’s projected revenue from ChatGPT is $2.7 billion this year, a nearly 300% increase from last year’s $700 million.

Then there’s Notion AI, which charges $8 more per month for access to its AI features that help users summarize notes and identify actions in a meeting. This integration gave the productivity app pride of place Forbes‘ sixth annual AI 50 list. Also on this AI 50 list are applications such as the $850 million automated medical documentation application, Abridge; Harvey, AI assistant for legal, tax and finance professionals, valued at $1.5 billion; and Synthesia, an AI avatar and video generator with a valuation of over $1 billion.

However, even with all these head-turning numbers, the operating costs of these platforms are substantial. Research and development costs for Nvidia’s Blackwell chip, for example, were around $10 billion, according to Huang.

Getting the right ROI on AI

As the need for AI chips and platforms increases, associated costs will likely increase, pushing companies to seek new revenue models or even adjust their prices to cover operational demands. This is especially true for AI-powered applications that interact directly with end users and need to keep pricing within reach to maintain a strong user base. They face the unique challenge of balancing affordability for customers with the high costs associated with AI processing and data management.

“We are still in the early days of monetizing AI-based applications, as people try to determine which AI-based workflows create tangible business value for customers. Regardless of how AI-based applications monetize, they often license AI as a service and pay real costs for it (in addition to paying real costs for the additional compute),” said Kyle Poyar, co-founder of Tremont, in an interview with Mainly metrics.

Added to this challenge is some uncertainty about the return on investment of AI-based solutions. As Medved explains: “There are real questions about ROI in AI, given that even trillion-dollar technology leaders are talking about a 15-year payback on investments. that they are currently carrying out. But they would rather take the risk of investing here than miss out on this unprecedented opportunity.”

For AI companies, the path to sustainable profitability will require navigating the tricky landscape of the technology’s high operational costs and the pressure to quickly demonstrate real value. The path to a sustainable business model may then depend on factors such as effective cost and price control.

As Poyar noted, “with their usage-based scaling costs, AI application providers may wish to charge their own customers based on usage and thus protect themselves against increasing their costs due to a few heavy users. »

The greatest pricing power, he said, will come from positioning AI as a way to increase revenue rather than just a time-saving solution. “This is because time savings are not differentiated, do not create urgency and do not bring in real money,” he added.

Another factor is prioritizing human capital to maximize returns on substantial investments.

“Spend your money first on hiring the best AI people possible and let them help you make those investment decisions.” Safe Superintelligence would have secured an initial investment of $1 billion at a valuation of $5 billion with just 10 employees, which would imply a value of almost $1 billion per employee,” Medved said.

Gilles Thonet, deputy secretary general of the International Electrotechnical Commission, admits he is not sure if the AI ​​bubble will burst, but says AI is largely embedded in our daily lives and that this does not not going to change.

“Risk is in the nature of startups: some succeed and unfortunately, others fail,” he said. “But by following global regulatory frameworks like the ISO/IEC AI standards can eliminate the need for businesses to start from scratch, guiding them to follow best practices in algorithm design and ensuring data quality to reduce AI bias.

Ultimately, business leaders who combine great talent with a strong vision for AI business applications are more likely to find themselves among the top 1% of AI companies that succeed and endure beyond of the current media hype.