SUMMARY

Liam Ottley, an AI agency expert, dissects the AI bubble driven by Big Tech's $400 billion spending, highlighting enterprise failures versus SMB successes, and offers a five-point playbook for AI agencies to thrive through 2026.

STATEMENTS

The AI industry faces a potential bubble due to more money being invested and spent on AI than earned from it, primarily fueled by Big Tech's $400 billion annual spending on data centers.
Circular spending among tech companies inflates revenues artificially, as firms pass money between each other without new funds entering the system, boosting stock prices based on misleading earnings.
The S&P 500's gains are heavily reliant on AI and tech stocks, making the broader economy vulnerable to a cascade failure if these stocks tumble due to unreturned investments.
Value from AI is expected in consumer tools like ChatGPT and business applications via custom AI systems, but adoption data reveals stark differences in success rates.
The dot-com bubble analogy shows telecom overbuilds led to unused infrastructure, but today's AI tools see massive consumer usage, unlike enterprise custom AI pilots.
An MIT report indicates a 95% failure rate for generative AI pilots in enterprises, particularly in custom development that requires rethinking processes for positive ROI.
A Wharton report counters with 75% of companies seeing positive ROI from AI, mainly from generic LLM tools like ChatGPT that boost employee productivity.
Smaller companies ($50-250 million revenue) achieve 79% positive ROI with AI, while multi-billion enterprises are three times more likely to stall in pilot phases.
AI automation agencies should target small to medium-sized businesses (SMBs), which are agile enough for transformational custom AI without enterprise bureaucracy.
Partnering with external vendors like AI agencies doubles the success rate of AI projects, as internal efforts in enterprises often fail without ongoing optimization.

IDEAS

Big Tech's massive infrastructure investments create a bubble illusion through circular revenue, where inter-company spending mimics growth without real economic input.
Consumer AI tools like ChatGPT demonstrate genuine value with hundreds of millions of active users, contrasting sharply with enterprise struggles in custom implementations.
The 95% enterprise failure rate stems from slow, bureaucratic structures unable to pivot to AI-first systems, opening doors for nimble SMBs to leapfrog competitors.
Generic AI tools provide quick productivity wins across all company sizes, but true transformation demands custom development that's easier for smaller firms to execute.
AI agencies can capitalize on enterprise pitfalls by focusing on SMBs, where agility allows for rapid rebuilding of processes with AI, yielding higher ROI.
Ongoing optimization, not one-time builds, is key to successful AI projects, involving feedback loops and iterations that agencies are uniquely positioned to provide.
Education and training on generic tools offer low-hanging fruit for ROI, serving as an entry point to deeper consulting and custom development services.
Retainers for essential AI systems ensure revenue stability during economic downturns, as clients prioritize critical operations over one-off projects.
The hype phase of AI adoption has ended, shifting to an era where proof of ROI is mandatory, favoring niched agencies with data-backed results.
External vendors like agencies can push custom AI success rates from 5% to higher by incorporating client learning and long-term feedback.

INSIGHTS

Circular spending in AI infrastructure masks a fragile market, but robust consumer adoption suggests the core technology will endure beyond hype-driven valuations.
Enterprise rigidity hampers AI transformation, revealing that scale can be a liability, while SMB agility positions them as future leaders in AI-driven efficiency.
Distinguishing generic tools from custom development clarifies adoption gaps: quick wins build momentum, but sustained value requires iterative partnership.
Focusing on ROI metrics transforms agencies from builders to strategic partners, aligning services with client demands in a maturing market.
Education as a gateway service democratizes AI benefits, bridging literacy gaps and paving the way for deeper implementations that yield compounding returns.
Retainer models foster resilience, embedding agencies into client operations to weather bubbles, ensuring mutual growth through economic cycles.

QUOTES

"There is more money being spent on AI and invested into it than is being earned from it."
"The 95% failure rate in these enterprises creates unprecedented opportunities for vendors who are able to build AI systems that incorporate learning and feedback from their clients."
"The smaller the company, the better the results. Firms in the $50 to $250 million per year range see a 79% positive ROI."
"Partnering with an external vendor like you and I doubles the success rate of the AI project."
"The real value for your clients is going to be unlocked by taking a very important system, kind of breaking it, putting AI into it, and then being there to help them through the process of optimizing over the long term."

HABITS

Maintain a sharp focus on SMB clients to leverage their agility for quick AI implementations and measurable results.
Routinely calculate and track ROI in every project, using client data to refine pitches and ensure tangible outcomes.
Prioritize ongoing optimization by scheduling regular feedback sessions and iterations on AI systems post-launch.
Integrate education and training as foundational services, starting with audits to assess team AI literacy before advancing to development.
Build retainer agreements into core offerings, emphasizing system criticality to secure recurring revenue streams.

FACTS

Big Tech companies are spending $400 billion annually on AI data center buildouts, driving much of the industry's investment.
75% of S&P 500 gains are attributed to AI and tech stocks, known as the Magnificent 7.
MIT reports a 95% failure rate for generative AI pilots in enterprises attempting custom development.
Wharton findings show 75% of companies achieve positive ROI from AI, rising to 79% for firms with $50-250 million in revenue.
Consumer AI tools like ChatGPT boast hundreds of millions of daily and weekly active users worldwide.
70-80% of employees in surveyed companies use AI tools weekly, indicating strong internal adoption.

REFERENCES

MIT report on generative AI pilots and 95% enterprise failure rate.
Wharton report on 75% positive ROI from AI investments.
ChatGPT as a leading consumer LLM tool.
Claude and other generic AI tools for business productivity.
Technology adoption lifecycle model, including early majority phase.
Dot-com bubble analogy involving telecom fiber overbuilds.

HOW TO APPLY

Avoid the enterprise trap by targeting small to medium-sized businesses, starting with very small clients and scaling to those with 100-500 employees for agile implementations.
Obsess over ROI by niching down to specific problems or industries, collecting performance data from past clients to provide data-backed estimates during sales.
Shift from builder to optimizer by designing projects with built-in feedback loops, conducting weekly tests and iterations to refine AI systems for reliability.
Become an AI transformation partner by leading with education, such as AI literacy audits and training workshops, to secure quick wins before progressing to consulting and development.
Default to retainers by integrating them into essential systems like sales funnels, ensuring clients view ongoing support as indispensable for operational continuity.

ONE-SENTENCE TAKEAWAY

AI agencies thrive by targeting agile SMBs, proving ROI through optimization and education amid the bubble's enterprise pitfalls.

RECOMMENDATIONS

Steer clear of large enterprises and double down on SMBs for faster, higher-ROI AI transformations.
Niche your services to gather concrete data, enabling undeniable proof of value in sales discussions.
Emphasize long-term optimization in contracts to evolve AI systems based on real-time client feedback.
Offer AI training and audits as entry-level services to build trust and upsell to custom development.
Structure deals around retainers for critical systems, safeguarding revenue during potential market downturns.

MEMO

In the frothy world of artificial intelligence, a bubble is forming—not from empty promises, but from a torrent of cash swirling among tech giants. Liam Ottley, founder of Morningside AI, warns that Big Tech's $400 billion annual splurge on data centers and GPUs is creating circular revenue streams, where companies like Microsoft and Google trade funds like hot potatoes, inflating stock prices without fresh economic value. This mirrors the dot-com era's overbuilt fiber optics, but with a twist: consumer tools like ChatGPT are already captivating hundreds of millions, proving AI's sticky appeal.

Yet, beneath the hype, enterprises are stumbling. An MIT study reveals a staggering 95% failure rate for custom AI pilots in these behemoths, where bureaucratic inertia prevents true reinvention of processes. Smaller firms, however, are sailing smoother waters. Wharton's research shows 75% of companies reaping positive returns, climbing to 79% for those with $50-250 million in revenue. Generic tools like Claude boost productivity effortlessly, while custom builds shine brightest among nimble SMBs unencumbered by legacy systems.

For AI agencies, this disparity is a clarion call. Ottley urges a pivot away from lumbering enterprises toward speedboat-like small businesses, where agencies can dismantle and rebuild operations from scratch. The playbook is clear: niche deeply to amass ROI data, then wield it like a scalpel in sales. Forget set-it-and-forget-it projects; become optimizers, iterating endlessly with client feedback to forge reliable AI agents.

Education emerges as the low-barrier gateway. Start with literacy audits—mapping team strengths in tools like ChatGPT—before escalating to workshops and bespoke development. This foot-in-the-door strategy yields quick wins, as Wharton's data affirms training's outsized impact. And to weather a pop? Lock in retainers for mission-critical systems, turning one-offs into indispensable lifelines that clients won't abandon in a downturn.

Ultimately, the bubble's shadow need not engulf everyone. Consumer frenzy and SMB agility signal AI's enduring power, not fleeting fad. Agencies that heed the data—focusing on proof, partnership, and persistence—stand poised not just to survive 2026, but to redefine industries in its wake. Ottley's optimism rings true: in this storm, the prepared will harvest generational wealth.