Temasek Holdings, one of Asia's largest investment firms, just announced it's doubling down on artificial intelligence by increasing its AI exposure from 6% to as much as 15% by March 31, 2031. This isn't a casual bet; it reflects a fundamental shift in how major institutional investors view the technology landscape. Rather than chasing the latest AI model or startup, Temasek is strategically positioning itself across the entire AI value chain, revealing which companies are likely to be the real beneficiaries of the AI boom.

What Is Temasek's AI Investment Strategy?

Temasek's CEO Dilhan Pillay explained that the company views AI as a "structural, long-term driver of value creation." This means the firm isn't treating AI as a temporary trend but as a fundamental shift that will reshape how businesses operate for decades. To capture this opportunity, Temasek is deploying capital across five distinct areas of the AI value chain:

Dilhan Pillay

• Energy and Data Centres: The physical infrastructure required to power AI systems and store massive amounts of data.
• Semiconductors: The chips that enable AI computation, from training to deployment.
• Cloud Services Providers: Platforms that deliver AI capabilities to businesses and consumers.
• Foundation Models: The large language models (LLMs) that form the basis of modern AI applications.
• AI Applications and Software Infrastructure: The tools and platforms that help businesses integrate AI into their operations.

Temasek is also specifically targeting "vertically integrated mega-caps," which are large companies with expertise spanning multiple areas of this value chain. This strategy reveals an important insight: the biggest winners may not be the companies building the flashiest AI models, but rather those providing the foundational tools and infrastructure that power the entire ecosystem.

Which Companies Are Positioned to Win?

NVIDIA stands out as perhaps the most critical player in this infrastructure race. The company is the world leader in graphics processing units (GPUs), which are the specialized chips required to train and deploy massive AI models. But NVIDIA's advantage goes far beyond just selling hardware. The company offers an entire ecosystem powered by its proprietary CUDA software platform, which makes it extremely costly for other tech giants or startups to switch to competing solutions.

The numbers tell a compelling story. In the first quarter ended April 26, 2026, NVIDIA's revenue surged 85% year-over-year to a record $81.6 billion. Its Data Centre segment, which serves AI companies and cloud providers, grew even faster at 92% to reach $75.2 billion in the same quarter.

"The extraordinary speed of the buildout of AI factories," explained CEO Jensen Huang, drove much of this growth.

Jensen Huang, CEO at NVIDIA

The new Blackwell architecture, NVIDIA's latest GPU design, was being massively ramped up to meet demand from companies building AI infrastructure.

What makes NVIDIA's position particularly resilient is that it doesn't matter which company wins the race to build the most advanced AI model. Whether OpenAI, Alphabet, or another frontier lab emerges as the leader, they all need NVIDIA's chips and software. As the "picks and shovels" provider to the entire AI gold rush, NVIDIA benefits from the competition itself rather than being threatened by it.

Taiwan Semiconductor Manufacturing Company (TSMC) represents another critical chokepoint in the AI supply chain. TSMC controls more than 90% of global advanced chip manufacturing, making it the only foundry currently mass-producing the leading-edge AI chips designed by NVIDIA and other major tech companies. In the second quarter of 2026, TSMC's revenue increased 33.7% year-over-year to $40.2 billion, while net profit climbed 77.4% year-over-year. Like NVIDIA, TSMC's business is expected to remain resilient regardless of which company dominates the AI application landscape.

Microsoft represents a different but equally important piece of the puzzle. While many people know Microsoft primarily for Office software like Word and Excel, the company is building what could become the ultimate AI ecosystem. Microsoft has already established itself as a leading cloud computing giant through Azure, secured a multi-billion-dollar stake in OpenAI, and embedded Copilot AI (its AI assistant) into its ubiquitous commercial software suite.

In the fourth quarter of fiscal year 2026, Microsoft's revenue increased 18% year-over-year to $90 billion, while diluted earnings per share rose 32% to $4.81. Azure and other cloud services revenue grew 43%, while paid Microsoft 365 Copilot seats surpassed 30 million, with net seat additions more than doubling quarter-over-quarter. Once Copilot is wired into enterprise users' daily workflows, the cost of switching to another provider becomes extremely high, creating what the industry calls "enterprise stickiness" that's nearly impossible to break.

How to Identify AI Winners Beyond the Hype?

For investors trying to navigate the AI boom, Temasek's strategy offers a practical framework for thinking about which companies are likely to thrive long-term:

• Look for Infrastructure Providers: Companies that provide the foundational tools, chips, or platforms that multiple AI companies depend on tend to be more resilient than those betting on a single AI model or application winning.
• Evaluate Switching Costs: Companies that create high switching costs through proprietary software platforms or ecosystem lock-in, like NVIDIA's CUDA or Microsoft's enterprise integration, have stronger competitive moats.
• Follow Real-World Productivity Gains: Focus on companies that are demonstrating actual productivity improvements and cost savings for their customers, not just theoretical potential.
• Consider Vertical Integration: Companies with expertise across multiple parts of the AI value chain, from hardware to software to applications, may have advantages over single-focus competitors.

The key insight from Temasek's strategy is that the AI era will likely produce winners across multiple layers of the technology stack. Rather than betting everything on which AI model or startup will dominate, smart investors are positioning themselves to benefit from the structural growth of AI infrastructure itself. This approach acknowledges that regardless of which specific AI application becomes dominant, the underlying infrastructure providers will continue to capture value from the entire ecosystem.

Microsoft's capital expenditure plans underscore just how serious the infrastructure race has become. The company is funneling its operating cash flow into infrastructure, with capital expenditure expected to reach roughly $175 billion in calendar year 2026, a figure revised down from about $190 billion due to an accounting change, not because the company is spending less. This massive investment signals that the race to build AI infrastructure is far from over and that the companies providing the picks and shovels will remain critical to the entire AI economy for years to come.