The conversation around Artificial Intelligence often centers on a few repetitive questions: Which model is the smartest? Which chip is the fastest? Which lab will reach the next breakthrough first?
As a tech commentator, I explore a central thesis in my book, AI Strategy That Wins: How Moats, Money and Business Models Decide the Next Decade: focusing strictly on technology misses the bigger picture. With vast enterprise spending flowing into generative AI, real success comes down to core strategy. When foundational intelligence becomes cheap and accessible to everyone, technology stops being the sole differentiator, strategy, finance, and business-model design become the real competitive advantages.
Inside AI Strategy That Wins
Originating as a series of essays written for his blog (blog.sharmavishal.com) and LinkedIn newsletter (Strategic Foresight), Sharma consolidates critical enterprise case studies, examining giants like Microsoft, NVIDIA, Amazon, Salesforce, Palantir, and Accenture, into a cohesive framework.
The book outlines Six Patterns of Advantage that separate long-term market leaders from fleeting tool providers:
Distribution: Owning direct routes to customers lets platforms deploy AI far faster than competitors can build alternatives.
Platforms: Converting products into sticky, interconnected ecosystems creates high switching costs.
Flywheels: Designing self-reinforcing loops ensures that early operational wins continually lower future costs.
Capital: Managing immense funding and infrastructure requirements allows companies to outlast long payback periods.
Pricing: Shifting monetisation away from static seat licenses toward outcomes and digital labor aligns price directly with delivered value.
Trust: Building auditable, reliable, and compliant AI agents ensures enterprise-grade deployment.
Rather than acting as a code tutorial or a set of speculative forecasts, the book delivers a practical playbook. It features a practical Six-Lens Framework designed to help business leaders, CIOs, and investors evaluate their competitive moats, financial stamina, and operational readiness.
Get Your Copy Today
Whether you are deciding where to allocate tech spend, rethinking your product pricing, or navigating the shift toward agentic workflows, AI Strategy That Wins offers a clear roadmap for the decade ahead.
I've just published a new book, From Mobile Towers to AI Tokens: How AI Could Save the Telecom Industry. It's a collection of my blog articles, presentations and other books since 2010, refreshed with 2026 results and brought together around one question.
In 2011 I gave a talk called Telecom: Endangered Species. Telcos were huge and essential, but growth had stalled and investors treated them like bonds. Fifteen years later, most of that still holds. What has changed is what the world now needs from them.
AI is short of power, land and sites close to users. Telcos already own powered, connected sites in almost every suburb and town. In October 2026, T-Mobile, Nokia and Nvidia completed what they called the world's first AI-RAN call on a live commercial network. The same machine that runs the radio can also run AI.
What the book covers
Telecom's story: how the industry was built, who the giants are and what their accounts show.
AI's appetite: its demand for power, and its own capital problem.
AI-RAN: a step-by-step financial model of a single site.
Winners and losers: from Australia to the world.
What next: nine risks, three possible futures and a practical playbook.
My view
I believe the opportunity is real, but smaller and slower than the headlines suggest. The first live call proved the radio works. It did not prove the economics. In my worked example, a site only breaks even when about 37% of its spare capacity is sold. It's a living book, so I will keep updating it as operators report. If you work in telecom, technology or investing, I'd welcome your challenge.
Download the eBook
📖 From Mobile Towers to AI Tokens: How AI Could Save the Telecom Industry.- Download here: Link
I'm pleased to share a release of my latest eBook at amazon, Four Boxes, Clear Thinking: The Art and Science of Matrices That Drive Decisions.
Almost every strategy deck has a two sided matrix in it. Some are brilliant: they turn a messy argument into two lines and four boxes and help a room agree on a direction in minutes. Many more are decoration, with axes that overlap, lines drawn by eye and boxes that never tell anyone what to do.
This book is about the difference. It is a practical guide to building your own matrices from first principles, so they hold up when an executive, a procurement panel or a board starts asking questions.
This book explores:
Why the two by two matrix has outlasted its critics, and the science of why four boxes fit the mind
A seven step method: question, axes, thresholds, quadrant names, plotting, testing and the so what
Twelve common traps and how to avoid them
Commercial uses including company analysis, organic growth, risk and return, account planning and sourcing
Technology uses including architecture, cloud, digital transformation and networks
Industry playbooks across healthcare, care services, financial services, government, telecom, energy, retail, manufacturing and education
When not to use a matrix, and how to use AI to build and stress test one
Every matrix has been drawn fresh for this book, drawing on more than 26 years of business and IT advisory work.
Whether you are an executive, consultant, sales leader, strategist or anyone who builds decision slides, I hope you find it useful.
This blog post examines the $1 trillion surge in AI investment as a strategically orchestrated system ofCircular Vertical Integration (CVI), where NVIDIA, Microsoft, OpenAI, Oracle, and AMD control the full technology stack while recycling capital through reciprocal investments. This closed-loop model has accelerated development by two years and reduced costs by 25%.
Drawing parallels with Tesla’s integrated strategy and the dot-com bubble, the analysis identifies an8 out of 10bubble risk—despite genuine revenue traction, such as OpenAI’s $4.5 billion in H1 2025. Key vulnerabilities include energy cost shocks, monetisation delays, and infrastructure bottlenecks.
The post highlights a potential30–50% market correction by mid-2026, driven by rising energy costs and lagging profitability. However, with a$10 trillion addressable market (TAM), long-term survivors could realise5× returns by 2030.
1. Building AI's Closed Kingdom: The Flywheel of Circular Vertical Integration (CVI)
The current $1 trillion surge in AI investment—driven by giants like NVIDIA, OpenAI, Microsoft, Oracle, and AMD—isn't a chaotic spending spree; it's a strategically synchronised operation. It’s what analysts are calling Circular Vertical Integration (CVI). This concept is the core mechanism enabling what is now being termed the "AI Flywheel."
1.1 What is Circular Vertical Integration (CVI)?
CVI is a structural and financial model designed to create a self-sustaining AI empire, locking out external competition and securing vital resources.
Vertical Integration (The Structure):This means controlling the entire supply chain, from raw computing power to the end-user application. Think of Apple designing its chips and software, or Ford historically owning steel mills. In AI, this means controlling the layers from chips (NVIDIA/AMD) to cloud infrastructure (Oracle/Microsoft) to the foundation models (OpenAI).
Circular (The Process):This is the financial loop. Initial investments aren't one-way tickets; they are recycled through reciprocal purchases, equity stakes, and R&D partnerships. For example, NVIDIA's investment in OpenAI is immediately recycled as OpenAI spends that capital on NVIDIA's chips, which then drives revenue back to NVIDIA for more R&D and investment.
Veteran analyst Jay Goldberg from Seaport Research Partners put it, this is "textbook vertical integration—a self-sustaining loop locking down the entire AI empire."
Source: Double Helix from Clockspeed- Vertical vs Horizontal Industry Structure
1.2 The AI Flywheel in Motion
This combination—CVI—creates a self-accelerating "flywheel,"a term famously used by Amazon for compounding momentum.
NVIDIA sells chips to Oracle and invests in OpenAI (via money or discounted hardware).
Oracle buys chips (like the $40 billion AMD order) to build its cloud capacity (OCI), which it then leases to model builders.
OpenAI uses this compute on Microsoft's Azure cloud (which also has a $13 billion equity stake) to train its models (GPT-5, Stargate).
Microsoft then embeds these models into its platform (Office, Azure), generating massive revenue and solidifying its moat of 1 billion+ users.
This revenue flows back to the cloud providers, chipmakers, and model developers, funding the next, faster spin of the flywheel.
Key AI Players - Vertical Play and Flywheel
This mechanism isn't about luxury; it's a competitive imperative. The$10 trillion compute requirements for futuremodels like GPT-5 mean these companies must secure their supply chains. Without this closed loop, companies risk being left behind. GPT-5's development was accelerated by two years, and costs were reduced by 25% through this model.
Jay Goldberg from Bloomberg says, "No circleequalsroadkill. GPT-5: 2 years faster, 25% cheaper."
AI Flywheel for Scale and Integration
2. The $1 Trillion Deal Web: Numbers and Interconnections
2.1 Circular AI Money Flow
The sheer scale of these interlocking deals justifies the "circular gamble" metaphor:
Circular AI Money Flow
AI Deal Web
This interconnected deal structure means that"half of global AI supply"is pre-sold or dedicated to this group by 2026.
Gene Munster of Deepwater Asset Management Money tweeted,"This isn't capex; it's a cartel. Twenty gigawatts—20 nuclear plants—pre-sold to themselves. Genius or insanity?"
2.2. Why They Have No Choice
Artificial Intelligence is no longer a strategic option—it’s a business imperative. With global AI investment forecast to reach$375 billion by 2025(UBS), the scale of opportunity is unprecedented. Yet, critical bottlenecks—compute, energy, and cost—threaten to stall progress and derail innovation.
Crisis to Advantage
Catherine Wood (ARK Invest) says:“Tesla 2.0. 80% of enterprise value will be AI by 2030.”—The AI arms race is underway. Only those who scale fast and smart will endure.
3. The Tesla & Google Playbook: Vertical Integration Precedent
The current AI strategy is a near-perfect mirror of the playbooks mastered by Tesla and Google—full-stack control as the only path to dominance.
3.1 The Tesla Parallel: Full-Stack Advantage Meets Existential Risk
Tesla’s success wasn’t just about electric vehicles—it was about owning the entire stack. From batteries and motors to software and charging infrastructure, Teslavertically integratedwhat others outsourced. Critics scoffed:"Car companies don’t make batteries."But Tesla proved that controlling every layer achieved tighter integration, faster iteration, and superior margins.
By the time legacy automakers caught on, Tesla had built a10-year lead. The lesson? Vertical integration is the path to exponential growth, but it requires surviving the initial capital-intensive period—what Elon Musk famously referred to as"production hell."
Source: Gen AI The New Reality - 2023
3.2 The Google Twist: Scale, Integration, and the Monetisation Gap
Google's AI strategy is equally ambitious. Through Gemini, it is embedding AI across its entire ecosystem:Search and YouTube (2 billion users), Workspace (3 billion users), and Android (3 billion devices).
CEO Sundar Pichai has made the strategy clear:“We’re prioritising growth over profits.”This is a classic land-grab—aiming to lock in users before monetising. This mirrors Tesla’s early years: scale first, then focus on margins. However, Google faces a severemonetisation lagin its AI cloud services, magnifying the core risk of the CVI model.
3.3 How AI Giants Are Following the Playbook
The AI ecosystem is now following this strategy to create commercially unviableswitching costsfor rivals:
Microsoft’s integration of OpenAI into the Office suite is analogous to Tesla embedding Autopilot into every vehicle. By making AI a native part of its productivity ecosystem, Microsoft significantly raises the switching costs for enterprises—retraining a billion users on an alternative platform becomes commercially unviable. This strategy reinforces the indispensability of the Microsoft stack.
Source: Gen AI The New Reality - 2023
3.4 The Risks: Circular, Capital-Intensive, and Volatile
Just as Tesla’s "production hell" nearly derailed the Model 3, the AI ambitions of Google and others could be slowed by infrastructure bottlenecks and economic headwinds. The key risks are familiar:
High Capital Intensity:The $1 trillion in capex is based on projected, not realised, profits.
Delayed Returns:Monetisation is lagging, especially in core AI cloud services.
Systemic Constraints:Energy and compute constraints could throttle the flywheel's spin.
The lesson is clear:vertical control creates moats, only if you survive the build phase.
4. AI Boom vs Dot-Com Bubble: Lessons and Differences
4.1 The AI Surge: Frothy, Fast, and Fundamentally Different
AI is undeniably in a bubble phase; however, there is some substance to it. Valuations havetripled in just two years, andcapital expenditure has surged to $344 billion—now 10× higher than revenue. Yet, this isn’t purely speculative: OpenAI alone generated$4.5 billion in revenue in H1 2025, proving that real cash is flowing.
AsJohn Addis ofIntelligent Investorobserves, today’s AI market resembles the dot-com era in terms of pace and hype—but not in fundamentals. WhileNVIDIA’s P/E ratio of 60×is high, it’s far more grounded than the200×multiples seen in 2000. Still, caution is warranted:MIT research indicates that 95% of AI projects currently deliver no ROI (return on investment), highlighting the risk of overreach."
AI Bubble Risk Breakdown
The key question isn't whether the AI market is a bubble; it'show advanced the bubble is, which analysts estimate at about8 out of 10. The most probable event to trigger a market correction would be a dual shock: a50% surge in energy costscoupled with the ongoing issue oflag in monetisation. This delay in monetisation is clearly visible in industry figures, such asOracle's AI cloud margins, which are currently only14%, lower than the71%achieved by its existing, broader cloud business.
Overall Risk Score: 8/10 – “Frothy”Most probable tipping point: energy cost shock by 2026.
Gene Munster says:
“This is a good bubble. With a $10 trillion total addressable market, there are trillions to be made by the survivors.”
The AI boom of 2025 is drawing inevitable comparisons to the dot-com bubble of the early 2000s. Back then,$1 trillion was invested in fibre infrastructure, of which70% remained unused, and theNasdaq plummeted by 78%. Today’s AI surge is similarly capital-intensive—but the underlying dynamics are far more robust.
John Addisfrom Intelligent Investor again offers a measured perspective:
“This is not an apocalypse. Expect a 30–50% correction, not a collapse.”
Comparison: Dot Com Vs AI in 2025
Summary: A Bubble with Substance and Survivors
Yes, a correction is likely—30–50% by mid-2026—but this is not a repeat of the dot-com collapse. The AI economy is underpinned byreal revenue, enterprise adoption, and infrastructure with demand. With a$10 trillion total addressable market, the survivors of this cycle could see5× returns by 2030.
5. The AI Horizon: What Could Go Right—And What Could Go Wrong
As the AI economy accelerates, the future remains deeply uncertain. While some see a trillion-dollar transformation unfolding, others warn of systemic risks that could derail progress. The truth likely lies somewhere in between. Below, we explore both sides of the equation—four optimistic scenariosthat could supercharge AI’s trajectory, andfive downside risksthat could trigger a major correction.
5.1 The Upside: Four Scenarios That Could Supercharge AI
Despite the froth, there are credible paths to exponential upside. If key breakthroughs align, AI could unlock trillions in value and reshape the global economy.
Optimistic Scenarios
These scenarios represent thebull casefor AI: a world where infrastructure scales, applications deliver, and the technology matures into a general-purpose force multiplier.
5.2 The Downside: Five Risks That Could Derail the AI Boom
While the upside is compelling, the risks are equally real—and in some cases, imminent. These scenarios could trigger a sharp correction or long-term stagnation.
Bear Case Scenario
These risks reflect thebear case: a scenario where capital intensity, regulatory pressure, and technical limits converge to stall momentum.
Conclusion: Choose Your Lens
Optimists see four powerful tailwinds. Pessimists see five systemic risks.The outcome depends on execution, timing, and resilience.
6. The Bottom Line
Circular Vertical Integration (CVI) represents more than a technological feat—it is a stroke of strategic brilliance.By consolidating control over chips, models, and infrastructure, AI leaders are effectively mitigating supply chain constraints, delivering significant cost savings (estimated at 25%), and locking in vast ecosystems of over a billion users behind formidable switching barriers.
Crucially, the revenue is genuine; figures like OpenAI’s $4.5 billion in the first half of 2025 demonstrate that this is a market driven by real transactions, not just hype.
However, the risks are equally material. With an estimated8/10 bubble risk, history serves as a clear reminder that even the most visionary concepts can buckle under the weight of market overbuild and delayed profitability.
Investors should anticipate a30–50% market correction by mid-2026, primarilytriggeredbyunforeseen energy price shocksandsustained lags in commercial monetisation.
For those firms and investors who successfully navigate this volatile period, the resulting rebound is projected to be profound, potentially delivering5× upside by 2030.
Note on Flywheel's Concept Popularised by Amazon
Source: Gen AI The New Reality - 2023
The Amazon Flywheel is a powerful, self-reinforcing business model that focuses relentlessly on thecustomer experienceto drive momentum and growth. The cycle begins with Amazon prioritising a superior customer experience by offeringLower Pricesand a vastSelectionof products. This draws significantTraffic(more customers) to the site. The increased volume and scale allow Amazon to lower itsCost Structure(through greater efficiency and fixed cost leverage), which, in turn, enables the company to offer evenLower Prices. This loop then fuels more growth, attracting both customers and more third-party sellers (increasing selection), spinning the flywheel faster and creating a sustainable competitive advantage. Crucially, the model emphasisesreinvestmentof profits back into the business—such as fulfilment centres and R&D—to continuously improve the customer experience and reduce friction, ensuring the flywheel keeps accelerating.
Next Post:UiPath's Course Correction in its Go-to-Market Strategy
