Since the launch of ChatGPT four years ago, the tech market has surged nearly 90%. While it’s tempting to label this a speculative bubble, the underlying numbers tell a vastly different story. Unlike the dot-com era, where valuations floated on hype and 100x multiples, today’s market rise is grounded in fundamental earnings. In fact, trading multiples are actually down roughly 20%, sitting comfortably below 20x forward earnings for the S&P 500.

High-tech equipment, software, and R&D now account for a staggering 55% of all US capital spending. We have officially entered a new technological epoch—the “Everything Cycle”—where AI is no longer just a software feature, but the primary driver of global industrial expansion.

1. The Age of Atoms: A Trillion-Dollar Infrastructure Buildout

The most profound shift AI has triggered isn’t in code—it’s in concrete, copper, and power grids. The digital revolution has thrust us back into the “Age of Atoms.”

To support the insatiable demand for compute, hyperscalers (Alphabet, Amazon, Meta, Microsoft, and Oracle) are channeling their massive near-term operating cash flows directly into physical infrastructure. Their combined capital expenditure is projected to hit $780 billion in 2026 and surpass $1 trillion annually by 2027.

This buildout recently surpassed the historic expansion of US railroads as a percentage of GDP. But it extends far beyond data centers and GPUs. It is spurring an economy-wide modernization effort requiring investments in:

• Power grids and cooling systems
• Water management and transit
• Heavy construction and skilled labor

Counterintuitively, this can benefit local communities. Because power grids are shared fixed costs, adding a massive, stable customer like a data center can actually lower residential electricity rates by spreading the fixed costs across more units of electricity.

Use this interactive model to explore how varying levels of AI capital expenditure and productivity gains might compound to impact broader economic growth over the next decade:

Key insight: The AI infrastructure boom is fundamentally an industrial boom. The companies that learn to master physical supply chains, negotiate energy deals, and finance heavy construction will have a massive competitive advantage.

2. Enterprise Adoption: The Power Law of AI

While the physical buildout is advancing at breakneck speed, actual enterprise integration remains surprisingly early.

On the surface, 69% of S&P 500 companies have “live deployments” of AI. However, if you look at the ultimate barometer—recurring, deeply influential workflows with quantifiable impact tracked over time—adoption sits at just 2%. Most enterprises are currently just scratching the surface with tools like Microsoft Copilot.

Because we are so early, a “power law” is emerging among early adopters:

• The top 1% of AI power users spend roughly 8x more on AI vendors than the top 10%.
• The most forward-leaning, AI-native companies are dedicating nearly 10% of their headcount budget to AI tools, compared to about 1% for standard Fortune 500 companies.

For the companies aggressively adopting these tools, the financial impacts are already measurable. Fintech platform Chime has reduced its cost-to-serve by over 10% year-over-year for four straight years (nearly a 50% cumulative reduction). Meanwhile, Shopify’s AI “Sidekick” has increased the percentage of new merchants reaching their crucial five-order retention milestone by 8%.

3. The Rise of Agents and Plunging Compute Costs

The next frontier of AI application isn’t just conversational chat; it’s autonomous agents executing multi-step workflows. Agentic token usage has grown 14x recently, driven by a critical underlying economic trend: the plunging cost of compute.

Thanks to fine-tuning, smaller specialized models, and smart routing (choosing the cheapest appropriate model for each sub-task), agent steps are becoming radically more affordable. For example, some financial chat workloads have become 10x cheaper to run.

This drop in cost is magical for software economics. It makes it practical for an AI agent to try a task, check its own work, and try again without becoming prohibitively expensive. As these costs drop by orders of magnitude, entirely new use cases—previously blocked by latency or expense—are suddenly unlocked.

4. Consumer AI: The Marketplace Shift

Consumer AI is in its infancy. Barely 2% of US households currently pay for an AI subscription. Yet, the users who do subscribe exhibit incredibly rare “smile retention curves”—meaning they don’t just stick around, their usage actually increases over time as the product improves.

This presents a looming paradigm shift for consumer discovery platforms and marketplaces. If consumers shift their intent-driven searches from Google or Amazon to persistent AI agents (like ChatGPT or emerging tools like Muse), the $70 billion highly-profitable retail advertising industry could be disrupted.

However, this doesn’t have to be a zero-sum game. Marketplaces like Instacart—where online grocery penetration is still early—stand to gain massive incremental Gross Merchandise Value (GMV) by seamlessly integrating with AI agents that automate user purchases, drastically reducing friction.

5. The Private Market Renaissance

The public software market has sharply bifurcated. Today, only 30% of public software companies are growing revenues at 20% or more. Companies that are perceived as AI “losers” are stalling, while incumbents that successfully integrate AI (like ServiceNow) and AI-adjacent sectors (like Cybersecurity and Observability) are thriving as agents access more internal systems.

The real explosive growth is happening behind closed doors. We are currently witnessing a “Private Market Renaissance.” AI related startups now account for 86% of US VC deal activity (up from 65% in 2025).

Furthermore, today’s late-stage private champions are reaching unprecedented scale before going public. The top six private AI companies (OpenAI, Anthropic, Databricks, Stripe, Wayve, Revolut) hold a combined valuation of roughly $2.4 Trillion—more than the combined market cap of all IPOs over the last decade (excluding SpaceX).

By staying private longer, these companies can take massive, long-duration bets on future technology without the immediate scrutiny of public quarterly earnings.

Looking to the Horizon

While Large Language Models currently dominate the headlines, the capital pouring into the “Everything Cycle” is funding the next wave of technological breakthroughs. Over the next five to ten years, this infrastructure will serve as the launchpad for:

• Robotics and Autonomy: Physical agents operating in the real world (e.g., Waymo scaling autonomous transport).
• AI x Bio: Radically accelerating drug discovery and hyper-personalized medicine.
• American Dynamism: Retooling defense technology, manufacturing, and national industrial capacity.

The AI boom is not just a software upgrade. It is a fundamental, physical rewiring of the economy that is just getting started.