{"id":176835,"date":"2026-06-15T08:00:00","date_gmt":"2026-06-15T06:00:00","guid":{"rendered":"https:\/\/www.startupbusiness.it\/the-economics-of-ai-prioritising-business-over-the-origin-of-models\/176835\/"},"modified":"2026-06-16T10:25:19","modified_gmt":"2026-06-16T08:25:19","slug":"the-economics-of-ai-prioritising-business-over-the-origin-of-models","status":"publish","type":"post","link":"https:\/\/www.startupbusiness.it\/en\/the-economics-of-ai-prioritising-business-over-the-origin-of-models\/176835\/","title":{"rendered":"The economics of AI: prioritising business over the origin of models"},"content":{"rendered":"\n

For much of the past decade, artificial intelligence has been discussed primarily through the lens of technological breakthroughs, increasingly powerful models and the race to achieve cutting-edge capabilities. Headlines have focused on record investments, ever-expanding computing requirements and the competitive positioning of technology leaders. However, as AI adoption moves from the experimental phase to enterprise deployment, a different set of questions is beginning to dominate conversations at executive level. <\/p>\n\n

How can organisations generate sustainable business value from AI? Which technologies offer the greatest return on investment? And how can companies balance performance, costs, governance and operational scalability? <\/p>\n\n

Increasingly, the answer depends less on where a model was developed and more on its ability to solve business problems efficiently and effectively. As noted by the Harvard Business Review, organisations are becoming more pragmatic in their AI strategies, focusing on deployment economics and operational outcomes rather than merely technological prestige (Joshi et al., 2025). <\/p>\n\n

This shift reflects the maturing of the AI market itself. During the early stages of generative AI adoption, organisations sought access to the most advanced models available. Today, however, AI is increasingly viewed as a business capability rather than a technological novelty. The conversation is shifting from model leadership to value creation. <\/p>\n\n

From model leadership to the economics of implementation<\/h2>\n\n

The early stages of the generative AI revolution were characterised by a race to build increasingly capable foundation models. Success was often measured by benchmark scores, the number of parameters and computational scale. Although these indicators remain relevant, they are no longer sufficient to guide corporate investment decisions. <\/p>\n\n

According to the Brookings Institution, different AI ecosystems have pursued distinct innovation strategies shaped by their respective technological, economic and regulatory contexts (Muro et al., 2026). As the adoption of AI expands into customer service, software development, workflow automation and decision-support functions, organisations are increasingly focusing on operational considerations such as implementation costs, governance requirements, scalability and long-term sustainability. <\/p>\n\n

This development is transforming the way in which technology investments are assessed. A marginal improvement in a model\u2019s performance may not justify substantially higher operating costs if comparable business outcomes can be achieved through more cost-effective alternatives. Consequently, organisations are placing greater emphasis on total cost of ownership and measurable business value (Joshi et al., 2025). <\/p>\n\n

The result is a broader shift from technology-driven decision-making to value-driven decision-making.<\/p>\n\n

The rise of multi-model strategies<\/h2>\n\n

One of the most significant developments in enterprise AI is the emergence of multi-model architectures.<\/p>\n\n

In much the same way as with the evolution of cloud computing, cybersecurity and enterprise software, organisations are increasingly seeking to avoid vendor lock-in. Instead, they are building portfolios of AI technologies tailored to different business requirements. <\/p>\n\n

Some models can be used to assist with programming and software development. Others can support document analysis, workflow automation, customer interactions or knowledge management. Rather than seeking a one-size-fits-all solution, organisations select technologies based on specific use cases and economic considerations (Joshi et al., 2025). <\/p>\n\n

This trend has been accelerated by the growing availability of open-weight models and increasingly diverse AI ecosystems. According to MERICS (2026), advances in the development of open AI have lowered the barriers to experimentation and customisation, enabling organisations to develop more flexible implementation strategies whilst reducing their reliance on proprietary platforms. <\/p>\n\n

Consequently, AI is increasingly becoming a portfolio management challenge rather than a technological decision where \u2018the winner takes all\u2019.<\/p>\n\n

Efficiency is becoming a strategic capability<\/h2>\n\n

Historically, technological leadership has often been associated with scale. In the field of artificial intelligence, this assumption has translated into larger datasets, greater computational resources and increasingly sophisticated foundational models. <\/p>\n\n

Recent developments suggest that efficiency could become just as important as scale.<\/p>\n\n

Various innovation ecosystems have responded to different constraints by focusing on optimisation, engineering efficiency and strategic implementation practices. According to research by the Brookings Institution, these different approaches demonstrate that significant innovation can emerge not only through increased computational resources, but also through better use of existing resources (Muro et al., 2026). <\/p>\n\n

For organisations considering investments in AI, this distinction is becoming increasingly important. The aim is rarely to implement the most sophisticated model available. Rather, the goal is to identify the solution that delivers the greatest business value relative to its cost, complexity and operational requirements. <\/p>\n\n

As the adoption of AI expands across various industrial sectors, efficiency is evolving from a technical consideration into a strategic capability.<\/p>\n\n

Infrastructure is becoming the new constraint<\/h2>\n\n

As organisations step up their AI roll-out, the focus is gradually shifting from the capabilities of the models to the infrastructure needed to support them.<\/p>\n\n

A recent Reuters report highlights how competition in AI is increasingly determined by access to computing resources, semiconductors, energy and digital infrastructure, rather than by the development of the models themselves (Reuters, 2026a). Similarly, discussions regarding emerging AI token markets and infrastructure investments suggest that the next phase of competition in AI could be decided as much by operational capacity as by algorithmic innovation (Reuters, 2026b). <\/p>\n\n

This development reinforces an increasingly obvious reality: AI is not simply software. It is an infrastructure. <\/p>\n\n

The ability to deploy AI at scale depends on access to computational resources, energy availability, data quality, cybersecurity controls, and governance mechanisms capable of managing risks and compliance requirements. As AI becomes integrated into business processes, these factors become critical determinants of competitive advantage. <\/p>\n\n

Market signals, narratives and perceptions<\/h2>\n\n

The rapid growth of the AI market has generated a significant amount of commentary regarding adoption patterns and competitive dynamics. Whilst some claims are backed by rigorous research, others stem from observations within the venture capital world, industry commentary and narratives from emerging markets. <\/p>\n\n

One of the most widely discussed examples is the claim that around 80% of US start-ups developing on open-source AI models use foundation models developed in China. This estimate has been widely attributed to remarks made by Martin Casado, a general partner at Andreessen Horowitz, and subsequently amplified by the tech media, venture capital discussions and social media commentary (36Kr, 2025; YouTube, 2026). <\/p>\n\n

It is important to emphasise that this figure should not be interpreted as a representative statistical survey of the entire US start-up ecosystem. No publicly available methodology, sample description or independent academic validation has been released. Consequently, the estimate should be understood as a reflection of trends perceived by the market, rather than as definitive proof of nationwide adoption. <\/p>\n\n

Nevertheless, the popularity of this statement is noteworthy in itself. It reflects a growing perception among investors and entrepreneurs that organisations are evaluating AI technologies on the basis of performance, cost-effectiveness and flexibility of implementation, rather than solely on the basis of the supplier\u2019s background. <\/p>\n\n

Several publicly reported examples support this broader trend. Airbnb CEO Brian Chesky acknowledged that the company has evaluated and implemented Alibaba\u2019s Qwen model in selected customer service applications, describing the technology as effective and cost-effective for specific use cases (Sircar, 2026). This example illustrates how organisations are increasingly willing to evaluate AI solutions from multiple ecosystems when this creates measurable business value. <\/p>\n\n

The growing interest in models such as DeepSeek, Qwen, Kimi, MiniMax and Yi-Lightning further highlights the increasing diversification of the global AI ecosystem. Rather than viewing these developments solely through the prism of geopolitical competition, many organisations are seeing them as evidence of a more competitive and innovative market, capable of offering a wider range of technological options. <\/p>\n\n

From technological competition to organisational capability<\/h2>\n\n

Perhaps the most significant shift currently taking place in the AI market is the move from competition centred on models to competition centred on capabilities.<\/p>\n\n

As access to advanced AI becomes increasingly widespread, the competitive advantage is shifting from ownership of the technology to organisational capability. The ability to govern, integrate, implement and continuously improve AI-powered workflows is becoming more important than access to any single model. <\/p>\n\n

Organisations that successfully combine technology, governance, workforce readiness and process re-engineering are likely to generate significantly greater value than those focused solely on technology adoption.<\/p>\n\n

The implications are far-reaching. Long-term success may depend less on choosing the \u2018best\u2019 model and more on developing the organisational capacity to assess, integrate, manage and implement AI effectively. <\/p>\n\n

In this context, AI ceases to be a purely technological challenge and becomes a challenge of organisational transformation.<\/p>\n\n

Conclusion<\/h2>\n\n

The AI market is entering a new phase characterised by greater diversity, increasing competition and an ever-stronger focus on economic sustainability.<\/p>\n\n

Organisations are moving beyond a narrow focus on the leadership aspect of the model, increasingly prioritising business value, operational efficiency, implementation flexibility, governance and organisational capability. This transition reflects the natural evolution of a maturing technology market. <\/p>\n\n

As AI capabilities become more accessible, the competitive advantage is shifting increasingly towards execution, organisational learning and strategic integration. The organisations that create the most value may not be those with access to the most advanced models, but those that develop the strongest capability to implement them effectively. <\/p>\n\n

Ultimately, the most important question is no longer which organisation develops the most advanced model. The more significant question is which organisations will be able to translate technological possibilities into sustainable business value most effectively (photo by Google DeepMind <\/a>on Unsplash<\/a>). <\/p>\n\n

References<\/strong><\/p>\n\n