Colin Stewart

Technology and business analyst focused on the economics of Silicon Valley and AI infrastructure investment.

Mentions

Appeared in

Discussed in

Key positions and views

The AI era will trigger a $10 trillion capital expenditure cycle, representing a tenfold increase over the entire cloud computing era.

Economic value and profitability have fundamentally shifted from the application software layer (dominant in the cloud era) to the foundational hardware layer, with chip providers like NVIDIA capturing the highest gross margins.

NVIDIA is a uniquely undervalued growth company, as its stock trades at a low earnings multiple (around 20x GAAP) relative to its massive projected growth rate of 40-70%.

The growth of the AI sector is physically constrained by significant, real-world supply chain bottlenecks, including semiconductor manufacturing at TSMC, electrical power availability, and a shortage of skilled labor.

AI is enabling a massive leap in productivity that allows companies to build products and maintain operations with significantly fewer employees, as exemplified by Block's 40% workforce reduction.

Podcast consensus on Stewart

Points of consensus

▶The transition to AI necessitates a massive capital expenditure cycle, estimated to be $10 trillion, a 10x increase over the cloud era's $1 trillion investment.Apr 2026

▶In the AI era, economic value and the highest gross margins are shifting from the application software layer to the foundational hardware layer, specifically to chip providers like NVIDIA and AMD.Apr 2026

▶The AI supply chain faces several critical choke points that constrain growth, including advanced silicon fabrication (TSMC), electrical power availability, and a shortage of skilled labor for data center construction.Apr 2026

▶NVIDIA is a pivotal company in the AI transition, and Stewart views it as undervalued, trading at just over 20 times GAAP earnings while projecting 40-70% growth.Apr 2026

Points of debate

▶Stewart's assertion that NVIDIA is one of the 'cheapest growth companies' is a debatable valuation claim, as a 20x earnings multiple can be considered high in other contexts, despite its significant growth projections.Apr 2026

▶The example of Block reducing its workforce by over 40% to leverage AI highlights a potential for major labor displacement, a contentious topic where the net effect on overall employment is widely debated.

▶Stewart's forecast of a $10 trillion AI CapEx cycle is contingent on overcoming the very supply chain choke points he identifies (power, labor, fabrication), creating a point of tension and uncertainty in his own thesis.Apr 2026

▶The long-term sustainability of value being concentrated at the hardware layer is debatable; as AI technology matures, value could migrate back to specialized models, applications, or data providers.Apr 2026

Key themes

▶The AI Capital Expenditure SupercycleApr 2026

Stewart posits that the shift to AI is not just a software evolution but a massive infrastructure build-out, requiring a 10x increase in capital investment over the cloud era, totaling an estimated $10 trillion. He specifies that hyperscalers alone will spend approximately $2.5 trillion by the end of 2027.

This frames the AI boom as a physical-world industrial revolution, suggesting that the most significant investment opportunities may lie in foundational infrastructure like hardware, data centers, and energy, rather than just software.

▶Value Chain InversionApr 2026

Stewart observes a fundamental reversal of where profits are made compared to the cloud era. While cloud saw high-margin software companies dominate, the AI era's best margins are found at the hardware layer with chip providers like NVIDIA and AMD.

Investors should re-evaluate traditional software-first investment theses, as the analysis suggests that foundational technology providers currently hold the most pricing power and profitability in the AI stack.

▶AI-Driven Productivity and Labor DisruptionApr 2026

Stewart highlights AI's rapidly advancing capabilities, particularly in coding, which he claims has jumped from a 'sixth-grade level to a PhD-student level' recently. He cites Jack Dorsey's Block reducing its workforce by over 40% as a prime example of companies using AI to operate with fewer employees.

This points toward a future of leaner corporate structures and significant labor market shifts, where AI enables substantial productivity gains but also raises questions about widespread job displacement.

▶Strategic Supply Chain BottlenecksApr 2026

Stewart emphasizes that the pace of AI adoption is limited by tangible, physical constraints. He identifies key choke points including TSMC's dominance in silicon fabrication, the availability of electrical power, and a shortage of skilled labor for data center construction.

The growth of the entire AI ecosystem is tethered to solving these physical-world challenges, making companies involved in energy infrastructure, advanced manufacturing, and specialized labor strategically critical.

Source episodes

Sentiment over time

Not enough data for timeline

Changes over time

Past: The Cloud Era

Characterized by a total capital expenditure of approximately $1 trillion, with value and high gross margins accruing primarily to application-layer software companies.

Circa 2022-2023

The 'ChatGPT moment' marks the start of the AI investment supercycle. In the 12-18 months that followed, Stewart notes that AI coding ability advanced from a 'sixth-grade level to a PhD-student level'.

Present (2023-2027)

Stewart describes this as the initial AI build-out phase, projecting $2.5 trillion in CapEx from hyperscalers and others. During this period, value is concentrated at the hardware layer.

Next 3-6 Months

Stewart anticipates a major catalyst with the release of new AI models trained on NVIDIA's Blackwell architecture, expecting a 'phenomenal step-up in performance'.

Future: The AI Era

Stewart forecasts the culmination of the current investment cycle, resulting in a total capital expenditure of approximately $10 trillion, fundamentally reshaping the technology landscape.

Suggested prompts

How does Colin Stewart's $10 trillion AI CapEx forecast break down across different segments of the technology stack, such as hardware, data centers, energy, and software? &nearr;What are the second-order effects of the value chain inversion Stewart describes, particularly for enterprise software companies and venture capital investment strategies? &nearr;Beyond the choke points Stewart identifies (TSMC, power, labor), what other potential constraints could derail the projected AI infrastructure build-out? &nearr;How does Stewart's assessment of AI's impact on labor at companies like Block reconcile with broader economic data on employment and productivity? &nearr;

Key concepts

NVIDIA 6 ep AI Capital Expenditure 5 ep Hyperscalers 3 ep Private Technology Markets 3 ep AI Coding Tools 3 ep Data Center Infrastructure 3 ep Value Chain Shift 2 ep AI Supply Chain 2 ep AI and Labor Productivity 2 ep US Industrial Policy 1 ep

Notable quotes

“And if you take the cloud in total, it's probably about a trillion dollars of CapEx. So that means the order of magnitude, one order of magnitude more is AI. That's a $10 trillion CapEx bill.”

Colin Stewart · The Architects of Value: Mark Edelstone and Colin Stewart on the Economics of Silicon Valley

“I mean, NVIDIA is probably one of the cheapest growth companies you can buy in the world. It trades at 20-ish times, a little over 20 times earnings, gap earnings, growing, you pick your number, anywhere between 40 to 70% in the next couple of years.”

Colin Stewart · The Architects of Value: Mark Edelstone and Colin Stewart on the Economics of Silicon Valley

“in the cloud world, the gross margins, many of the gross margins, the best gross margins were actually in the application layer, right? So the biggest software companies were earning 80 85 gross margins in that business And as you move to AI actually some of the best gross margins are actually at the chip provider level of that technology It the NVIDIAs it the AMDs of the world”

Colin Stewart · The Architects of Value: Mark Edelstone and Colin Stewart on the Economics of Silicon Valley

“TSMC is a choke point, right? That all the silicon has to go through, TSMC mostly... You have power. The availability of power in the US is an issue... You have the lack of infrastructure from a workforce perspective to be able to put these together.”

Colin Stewart · The Architects of Value: Mark Edelstone and Colin Stewart on the Economics of Silicon Valley

Report last updated: Apr 21, 2026

Get started free

Back to Entities Intelligence Report