Business integration of AI is likely to be characterized by stops and starts, as organizations grapple with the complexities of implementation and change management.
In brief
- The rapid growth in AI applications is driving substantial spending on software, hardware and cloud resources, presenting opportunities for technology providers, but also raising concerns about infrastructure bottlenecks and rising costs.
- AI is set to disrupt traditional business models and reshape industries, with the U.S. and China taking different approaches to development and monetization, and the emergence of physical AI (such as autonomous vehicles and robots) further expanding the technology’s impact.
- Not all AI models and hyperscalers will come out on top; investors could start to differentiate those with more promising prospects, which will require active management.
There is a healthy level of skepticism regarding the development of artificial intelligence (AI) in the U.S. Investors are asking questions such as “Is there a strong enough use case?”, “Are AI hyperscalers over-investing in data centers and infrastructure?” and “Is the current valuation already reflecting much of the medium-term growth in this industry?”
Meanwhile, Asian exports have been well supported by demand for semiconductors and memory chips to facilitate this development. China is also pushing ahead with its AI development, despite hurdles such as U.S. export restrictions on high-powered graphics processing units (GPUs) to train and operate AI models.
Still plenty of room for demand to grow
We are comfortable with the potential demand for AI adoption and applications. Current usage data and business surveys all point toward rising adoption. This momentum is expected to continue as new applications and capabilities come online, although the path forward will not be entirely smooth. Business integration of AI is likely to be characterized by stops and starts, as organizations grapple with the complexities of implementation and change management.
The financial implications of this adoption wave are significant. Today, large enterprises spend approximately USD 3,500 per employee each year on software and SaaS solutions. With advanced AI services priced at USD 200 per month, it is reasonable to expect that businesses will be willing to pay for a suite of AI-powered tools, driving robust monetization across the sector. However, as AI models are tasked with increasingly complex, real-world problems, the computational requirements are rising significantly. Reasoning inference models, for example, may require up to 100 times more computing power than the single-shot models of previous years.
This surge in demand for computing resources presents both an opportunity for hardware and cloud providers and a risk in terms of potential infrastructure bottlenecks and rising costs. This trend benefits hardware and semiconductor manufacturers in both the U.S. and APAC.
Starting to monetize products and applications
From a monetization perspective, AI-native companies are demonstrating economic agility. AI model developers are already generating revenue in the billions of dollars. Although this is a fraction of the hundreds of billions that hyperscalers are pouring into capex, the general expectation is that this revenue stream will grow rapidly in the next 3–5 years, especially if competition drives out less competitive products. This will also demand deeper penetration into enterprise markets, expansion of premium subscriber bases and the development of new monetization channels such as advertising, business services and in-app purchases. This evolution is likely to disrupt incumbent business models, forcing established tech platforms to adapt or risk obsolescence.
Beyond that, the emergence of physical AI, such as autonomous vehicles and humanoid robots, further illustrates the transformative potential of the technology. These machines leverage generative AI to process sensor and video data in real time, enabling end-to-end intelligence and autonomous decision-making. Yet, this sophistication comes at a cost, requiring significantly more electricity and infrastructure than traditional electric vehicles—a trend that is likely to extend to AI-powered robotics and other physical applications.
Despite the attractive opportunities, investors should remain mindful of the risks. The uneven pace of business adoption, potential infrastructure bottlenecks, margin compression and the threat of incumbent disruption all pose challenges. Nevertheless, the willingness of enterprises and consumers to pay for advanced AI tools in the U.S. is rising, and companies that can scale infrastructure, capture demand and adapt to the evolving ecosystem are well positioned to deliver outsized returns.
China’s AI development could look somewhat different to the U.S. Its hardware limitations could force AI model developers to focus on more solution-specific model development. The less prevalent subscription-based model would also force local hyperscalers and technology conglomerates to focus on embedding AI into their existing services, such as financial service, e-commerce and other lifestyle services. More sector-specific AI development could be carried out for areas such as health care, professional services and manufacturing. Similar to the U.S., integrating AI into autonomous vehicles and humanoid robots would be the next extension of AI development, and potentially be a new source of exports.