NextFin News - As of February 16, 2026, the global technology landscape is witnessing a historic reallocation of capital that is redrawing the hierarchy of the world’s most valuable companies. According to data from RBC Wealth Management, the "Big Five" tech giants—Amazon, Alphabet, Microsoft, Meta, and Oracle—are projected to collectively spend approximately $650 billion on artificial intelligence infrastructure this year alone, representing a staggering 60% increase year-over-year. This unprecedented surge in capital expenditure is no longer just a speculative bet; it is a physical reality manifested in data center permits, power contracts, and massive supply agreements that favor the architects of the AI era.
While Amazon continues to leverage its cloud dominance through AWS, the sheer velocity of growth in the semiconductor and networking sectors has led analysts to predict a major reshuffling of the market cap leaderboard. According to reports from The Globe and Mail, two specific companies—Nvidia and Broadcom—are currently positioned on a trajectory that could see them exceed Amazon’s total market valuation within the next three years. This shift is driven by a fundamental change in how technology is built, moving away from general-purpose cloud computing toward specialized "AI factories" that require the high-density hardware these two firms provide.
Nvidia, led by CEO Jensen Huang, has successfully transitioned from a chip designer to a full-stack infrastructure provider. Huang recently described the current period as the "largest infrastructure buildout in human history," with global AI spending expected to reach $4 trillion by 2030. The company’s strategic pivot in 2026 involves selling entire rack-scale solutions, such as the GB200 NVL72 systems, rather than individual GPUs. This move significantly increases average selling prices and solidifies Nvidia’s grip on the data center ecosystem. With the new Vera Rubin architecture entering full production in the second half of 2026, Nvidia has already secured cumulative revenue visibility exceeding $500 billion for the 2025-2026 period.
The second contender, Broadcom, is capturing the rapidly expanding market for custom silicon. As hyperscalers like Google and Meta seek to reduce their reliance on off-the-shelf parts, they are turning to Broadcom to co-develop custom AI accelerators. According to research firm Counterpoint, Broadcom is estimated to command nearly 60% of the custom AI chip market by 2027. The company entered 2026 with a consolidated backlog of $162 billion, of which $73 billion is specifically tied to custom AI projects. This massive backlog, combined with a 65% year-over-year jump in AI-related revenue, provides a level of fiscal predictability that few other tech firms can match.
The analytical framework for this prediction rests on the concept of "capital intensity." For over a decade, Amazon and its peers enjoyed premium valuations based on asset-light, software-driven models. However, the AI era has forced a return to asset-heavy infrastructure. While this introduces new risks, such as technological obsolescence and higher depreciation, it also creates a massive "toll-booth" effect for the companies providing the essential components. Nvidia’s gross margins, currently maintained in the mid-70% range despite production scaling, demonstrate a level of pricing power that historically precedes massive market cap expansions.
Furthermore, the valuation gap is narrowing. While Amazon’s retail and advertising arms provide a stable floor, they do not offer the "parabolic" upside seen in the semiconductor sector. As of early 2026, Nvidia’s forward P/E ratio has actually moderated to its lowest levels since 2015 relative to its growth rate, suggesting that the market has yet to fully price in the long-term recurring revenue from its software and networking stacks. If Nvidia and Broadcom continue to capture the lion's share of the $650 billion annual spend from their own customers, the mathematical path to surpassing Amazon’s valuation becomes increasingly clear.
Looking forward to 2029, the primary risk to this prediction remains a potential "capex hangover" if the monetization of AI applications fails to materialize for the end-users. However, as long as the arms race between U.S. President Trump’s domestic tech initiatives and global competitors continues, the demand for the underlying hardware remains inelastic. For investors, the next three years will likely be defined not by who owns the cloud, but by who owns the silicon that powers it.
Explore more exclusive insights at nextfin.ai.
