Opinion
Why are AI providers lowering prices even as usage expands?
"For investors, the trend may look like a decline in profitability," writes Tzvika Ben Avraham, Partner, Head of Economic Advisory at Deloitte Israel, "but it may also reflect a combination of technological efficiency, capital investments, and competition for market share."
Why does the AI market challenge valuation models, and what can investors learn from this? In traditional capital markets, it is generally assumed that when the demand for a product or service increases, the price tends to rise accordingly. In the artificial intelligence industry, a more complex picture has been unfolding in recent times: the use of AI applications is expanding, while at the same time, some of the leading players - including OpenAI and cloud infrastructure providers - are reducing prices for certain services and models. In some cases, reductions of up to 80% in the price of certain models have been reported in recent months. (According to OpenAI's announcement, the company reduced the price of GPT-5.6 Luna by 80%, lowered the price of GPT-5.6 Terra by 20%, and improved the performance of GPT-5.6 Sol via API while leaving its price unchanged, effective July 30.)
For those examining companies primarily through their profit and loss statements, this trend may raise questions. On the one hand, companies in the sector are planning significant capital investments in setting up data centers and purchasing chips and infrastructure. On the other hand, price reductions could pressure profit margins and short-term cash flow. At first glance, this complicates the use of traditional valuation models. However, a broader analysis indicates that the trend may reflect several concurrent economic mechanisms rather than merely price erosion.
The first economic logic relates to the cost structure. Investments in infrastructure are largely upfront capital expenditures designed to increase capacity and improve efficiency. As computing power, software, and operational processes improve, the cost per unit of use may decline. Therefore, examining profitability cannot rely solely on current quarterly results; it must also account for the evolution of service costs over time, infrastructure utilization rates, and the scale of future demand.
The second logic is tied to the impact of price on the volume of usage. When technology becomes cheaper and more efficient, it can enable organizations to expand existing uses and adopt new ones. A significant drop in AI prices can improve the economic feasibility of compute-intensive applications, including digital agents and complex automated processes. However, the extent of demand growth and the ability to translate it into revenues and profitability depend, among other things, on product quality, the intensity of competition, the pricing structure, and the actual adoption rate.
The third logic touches on financial resilience and economies of scale. Companies with high access to capital and the ability to finance prolonged investments may be in a better position to absorb periods of low profitability, expand infrastructure, and compete on price. In addition, major companies in the field are subject to significant pressures to demonstrate rapid growth and justify high market valuations, and therefore they also act to capture market share at a relatively early stage and build a customer base to grow alongside. Alongside this, financial advantage does not guarantee success on its own: the outcome also depends on innovation, execution, customer trust, regulation, and the ability to generate value over time.
From an investor's perspective, the conclusion is not that price drops necessarily indicate weakness or, conversely, a successful takeover maneuver. It is more accurate to examine a combination of metrics: cost per unit of use, the growth rate of consumption volumes, infrastructure utilization, customer retention, technological differentiation, the scale of capital investments, and the ability to convert demand into sustainable cash flow. In a rapidly evolving market, a valuation model based solely on immediate profitability may miss part of the picture. At the same time, expectations for future growth must also be carefully scrutinized and supported by data.
The information in this article is for general information purposes only and does not constitute investment advice, a recommendation, or a substitute for professional advice tailored to the circumstances of any individual case.
Tzvika Ben Avraham is Partner, Head of Economic Advisory at Deloitte Israel
