Token Mania Is Fading: Samer Choucair Identifies the Sectors Benefiting from AI Efficiency
Investment leader Samer Choucair said the artificial intelligence sector has entered a phase of structural transformation as the phenomenon of excessive “token” consumption that accompanied the initial wave of AI enthusiasm begins to fade.
He explained that major companies have started imposing stricter controls on spending after AI budgets exceeded targeted levels.
Choucair noted that developers are increasingly turning toward smaller and open-source models in an effort to achieve greater efficiency at lower cost. He emphasized that this shift is affecting capital flows, technology-company valuations, and asset-allocation strategies across global and Gulf markets, alongside the acceleration of Saudi investment under Vision 2030.
Choucair explained that a significant proportion of executives exceeded their AI budgets over the past year, prompting companies such as Uber to set a monthly employee spending cap of $1,500, while Tesla established a weekly limit of $200. Gusto, meanwhile, adopted flexible limits after successfully reducing token costs by 13% through adjustments to its use of advanced models.
The End of the Unlimited-Spending Era
Samer Choucair said the initial wave of enthusiasm encouraged companies to use AI tools without clear spending constraints, turning bills for advanced models into an increasingly significant financial burden.
Executives eventually recognized that higher consumption did not necessarily translate into greater productivity or measurable returns.
Choucair added that companies rapidly reassessed their spending patterns, making AI-cost control an integral part of technology-project governance.
The shift, he said, reflects the market’s transition from an expensive phase of experimentation and exploration to one focused on efficiency and returns.
He noted that falling inference costs in recent years, driven by algorithmic improvements and intense competition—particularly from open-source models—have enabled companies in many cases to achieve comparable performance at lower cost.
This has encouraged developers and companies to adopt smaller, more specialized models capable of operating more efficiently on available infrastructure, whether on-premises or through cloud computing.
Reallocating Capital in AI
Samer Choucair emphasized that the shift reflects a maturing technology-adoption cycle, as markets move from costly experimentation toward structural optimization.
Institutional investors, he said, are increasingly monitoring how companies are redirecting budgets away from indiscriminate consumption and toward efficient infrastructure and specialized models.
Choucair said this transformation could reshape profit margins across the technology sector, particularly as competition shifts from the ability to provide the greatest amount of computing power to the ability to extract the highest possible value from each unit of computing capacity.
He explained that companies adopting model routing based on task requirements have an opportunity to generate significant savings by using smaller models for routine tasks and reserving advanced models for complex workloads while maintaining the required quality levels.
Infrastructure and Open-Source Models
Samer Choucair noted that infrastructure providers focused on energy and computing efficiency have continued to benefit from sustained demand for data centers, even as the cost per token has declined.
He added that open-source models have gained momentum as a tool for more efficient capital allocation, particularly in environments requiring data sovereignty or high levels of local customization.
This, in turn, has supported the growth of open-source software ecosystems as well as deployment and optimization tools.
Choucair said the current phase has prompted institutional investors to reassess their AI portfolios, with capital gradually moving toward companies that have demonstrated an ability to generate sustainable returns through operational efficiency rather than through consumption volume alone.
Saudi Arabia Bets on Digital Infrastructure Efficiency
In Gulf markets, Samer Choucair explained that this transformation has coincided with major investments in digital infrastructure under Saudi Vision 2030.
He noted that the Public Investment Fund, through initiatives such as HUMAIN and Alat, is working to build sovereign AI capabilities, including advanced data centers and Arabic-language large language models.
Choucair added that partnerships with global providers to strengthen computing capabilities have reinforced Saudi Arabia’s position as a regional AI hub.
Investment in efficient and open-source models, he said, could reduce operating costs and accelerate the adoption of digital solutions across non-oil sectors.
He emphasized that this approach allows Saudi Arabia to combine infrastructure investment with greater control over usage costs, consistent with the economic-diversification objectives of Vision 2030.
New Opportunities and Risks to Profit Margins
Samer Choucair pointed to new investment opportunities in specialized software, inference-management tools, edge computing, and solutions tailored to sectors such as financial services, healthcare, and logistics.
He added that energy and infrastructure providers have continued to benefit from sustained demand for computing, although that demand is becoming increasingly tied to efficiency.
Meanwhile, providers of advanced models could face pressure on their margins if the rapid shift toward lower-cost alternatives continues.
Choucair also noted that governance and compliance represent additional challenges in environments relying heavily on open model usage, particularly as AI applications expand within enterprises.
Computing Demand Will Continue—but Spending Is Changing
Samer Choucair said total demand for computing is likely to continue growing as intelligent agents and advanced applications become more widespread.
However, he emphasized that the distribution of spending will change significantly in favor of more efficient solutions.
Choucair stressed the importance of directing long-term investment toward building genuine productive capacity, whether through developing human skills or establishing digital infrastructure capable of supporting new technologies efficiently.
He added that markets capable of combining technological efficiency with a coherent national strategy—as in the case of the Saudi economy—will be better positioned to achieve a sustainable competitive advantage in capital allocation.
A New Era of Capital Allocation
Samer Choucair concluded that the second half of the decade could see a further decline in the era of uncontrolled token consumption, giving way to a more disciplined phase focused on the actual return on AI investment.
He explained that institutional investors, sovereign wealth funds, and asset managers will increasingly monitor companies capable of turning this transformation into a competitive advantage by improving business models and reducing structural costs.
Choucair emphasized that Gulf economies—and Saudi Arabia in particular—have an opportunity to expand investment in the digital economy as part of their diversification strategies.
Combining efficient and open-source models with advanced infrastructure, he said, could increase AI’s contribution to GDP while maintaining greater control over costs.
He stressed that strategic capital allocation, sound governance, and investment in long-term capabilities will remain the decisive factors in transforming the technological shift in artificial intelligence into sustainable economic value.