Is the AI Investment Bubble Going to Burst? Tech Industry Analysis
Problem
I’ve been watching a strange paradox unfold. On Reddit, someone captured it perfectly:
“Linus said it best that this tech is almost certainly going to change every industry and it’s also almost certainly hype.”
Both statements are true. AI will transform everything. AI valuations are also inflated beyond reason. I needed to understand how both could coexist—and when the bubble would actually burst.
The Core Question
Is the current AI hype sustainable?
I spent weeks researching this. The answer is clearer than I expected: yes, the bubble will burst, but the technology isn’t fake.
The distinction matters. The dot-com bubble burst, but the internet still changed the world. Housing bubbles burst, but houses still exist. AI will follow the same pattern.
What I Found
1. The Paradox Is Real
I dug through tech forums and found this insight that scored highest in community discussions:
THE AI PARADOX MODEL=====================
Technology Reality Investment Reality ----------------- ------------------ Transformative capability Irrational exuberance Genuine productivity gains 80x revenue multiples Measurable enterprise ROI Unprofitable growth Decreasing inference costs Massive burn rates
These are BOTH true simultaneously.One commenter explained why this matters:
“We are in weird times in tech because we’re in an investment bubble that is out of control and the new hot tool has been engineered to take advantage of our quirky hyper-social monkey brains to appear more lifelike and human while being manipulative.”
The bubble isn’t just about money—it’s about psychology. AI feels more transformative than previous hype cycles because it mimics human interaction.
2. The Burst Mechanism
Here’s what will trigger the collapse:
THE BUBBLE BURST MECHANISM===========================
Phase 1: Growth-at-all-costs (CURRENT) - Free or subsidized access - Massive venture capital inflows - User acquisition over revenue - $11M/month burn on $3M revenue
Phase 2: Profitability Pressure (NEXT) - Investors demand returns - Monetization experiments begin - Free tiers degrade - API prices increase
Phase 3: Enshittification (THE TRIGGER) - Quality degrades to extract value - User experience suffers - Customers churn - Revenue reality check
Phase 4: Market Correction (THE BURST) - Unprofitable companies fail - Consolidation accelerates - Valuations reset to fundamentals - Survivors emerge strongerOne Reddit comment nailed the trigger:
“Fortunately, this bubble will burst when these companies attempt to become profitable and greedy corporate decision makers will decide they need to monetize and therefore enshittify it.”
The burst won’t come from technology failure. It will come from the impossible math of profitability.
3. Historical Context: Microsoft FrontPage Comparison
I researched previous “revolutionary” tools. The Microsoft FrontPage story is instructive:
HISTORICAL "CODE GENERATOR" COMPARISON======================================
+----------------+-------------------------+------------------------+| Tool | Promise | Reality |+----------------+-------------------------+------------------------+| FrontPage | "Grandmothers become | Crappy websites, || (1996-2003) | web developers" | discontinued |+----------------+-------------------------+------------------------+| WebMatrix | "Democratize web | Limited adoption, || (2010-2012) | development" | discontinued |+----------------+-------------------------+------------------------+| Expression Web | "Professional design | Discontinued 2012, || (2006-2012) | for everyone" | replaced by VS |+----------------+-------------------------+------------------------+| AI Code Tools | "Replace developers" | Jury still out || (2023-?) | | |+----------------+-------------------------+------------------------+
Pattern: Tools promising to eliminate expertise produce mediocre outputs and fail.One forum poster noted:
“We’ve had ‘code generators’ since forever, and it never worked: Microsoft Frontpage, Microsoft Webmatrix, Microsoft Expression… and the AI slop is not any different.”
But here’s the critical difference I found:
WHY AI MIGHT BE DIFFERENT==========================
Static Tools (FrontPage): - Templates and wizards - No learning capability - Fixed output quality - Cannot handle complexity
AI Tools (LLMs): - Learned from billions of examples - Can improve over time - Handles genuine complexity - Integrates into workflows
WHY AI MIGHT BE SAME====================
Current outputs require expert correctionQuality varies dramatically by use caseVendor lock-in risksEnshittification inevitabilityI think the truth is in between. AI is fundamentally more capable than FrontPage. But the investment bubble around it is similar to every previous hype cycle.
4. The Enshittification Cycle
I studied Cory Doctorow’s enshittification theory. It predicts exactly what will happen:
THE ENSHITTIFICATION PREDICTIVE MODEL=====================================
Current State (2024-2026): - Free Claude, ChatGPT tiers available - API costs subsidized by venture funding - Features added rapidly - Users acquired with minimal friction
Near Future (2026-2027): - Free tiers restricted - API costs increase 3-5x - Features gated behind paywalls - Quality degradation to cut costs
Post-Bubble (2027-?): - Only paying customers survive - Enterprise pricing dominates - Consumer AI becomes luxury - Open source alternatives fill gaps
Pattern observed: Twitter, Facebook, Amazon, Uber, TikTok All followed this exact trajectory.The historical precedent is overwhelming. Every platform that starts free and subsidized eventually enshittifies when profitability pressure arrives.
Why This Matters
For Tech Workers
I analyzed career implications:
CAREER DECISION MATRIX======================
HIGH TRANSFERABILITY SKILLS: - Prompt engineering -> Clear communication - RAG systems -> Information retrieval fundamentals - LLM integration -> API design patterns - AI product work -> General product skills
LOW TRANSFERABILITY SKILLS: - Specific model optimization techniques - Proprietary framework expertise - Training pipeline work (unless infra-focused) - AI startup-specific knowledge
Decision framework: - Build skills that survive bubble burst - Don't assume AI salaries stay inflated - Maintain fundamentals while learning AIFor Investors
The investment framework is straightforward:
INVESTMENT SURVIVAL QUESTIONS=============================
1. Revenue > $50M? [YES/NO]2. Revenue growth > 100% YoY? [YES/NO]3. Burn rate < 3x revenue? [YES/NO]4. Clear competitive moat? [YES/NO]5. 2-year funding winter survival? [YES/NO]
Score 4-5: Consider investmentScore 2-3: Proceed with cautionScore 0-1: Avoid
Most hyped AI startups score 1-2.Infrastructure companies score higher.For Businesses
Practical advice for companies adopting AI:
DO THIS: - Start with specific, measurable use cases - Calculate ROI per project before scaling - Build internal expertise gradually - Avoid vendor lock-in where possible - Budget for price increases
NOT THIS: - "AI-first" transformations - Massive consulting engagements - Betting critical paths on AI - Assuming current pricing persists - Long-term contracts at current ratesKey Metrics I’m Watching
I’ve developed a tracking system for bubble signals:
BUBBLE COLLAPSE SIGNALS=======================
IMMEDIATE WATCH: - OpenAI/Anthropic pricing announcements - Free tier restriction announcements - AI company layoff announcements - Down round funding news
MONTHLY TRACKING: - Enterprise AI renewal rates - API pricing trends (per-token costs) - AI engineer salary normalization - AI startup closure announcements
QUARTERLY ANALYSIS: - Venture funding to AI startups - Public AI company valuations - Enterprise AI adoption surveys - Energy consumption trends
Signal strength: 2/10 (mild Phase 1 signs emerging)Expected peak: Late 2026 to early 2027The Environmental Factor
I can’t ignore this. One commenter noted:
“Is AI terrible for society (and the environment)? Yes. Will rich people push forward with it no matter the costs? Yes.”
AI ENVIRONMENTAL COSTS======================
Training single large model: ~50 GWhEquivalent to: 5,000 homes for a year
Daily ChatGPT queries: ~100M+Energy per query: ~0.3 WhDaily consumption: ~30 MWh
Annual AI compute growth: ~10x
Hidden costs: - Water for data center cooling - Carbon emissions from compute - E-waste from rapid hardware turnover - Land use for new data centers
Bubbles ignore externalities.Regulation eventually catches up.My Mental Model
I’ve developed a framework that helps me think clearly about AI:
THE AI DUALITY FRAMEWORK========================
TECHNOLOGY ECONOMICS ---------- ---------Reality: Transformative OvervaluedTimeline: 10+ years 1-2 year correctionRisk: Ignoring it Betting everythingOpportunity: Build skills Avoid FOMORight move: Learn & experiment Stay diversified
Both columns are true simultaneously.This is not a contradiction.This is how bubbles work.Common Mistakes I See
I watch people make these errors repeatedly:
1. All-or-Nothing Thinking
“I need to go all-in on AI or get left behind.”
Reality: Incremental adoption with measured outcomes beats FOMO-driven pivots every time.
2. Conflating Technology with Companies
“AI is revolutionary, so this AI startup will succeed.”
Reality: Revolutionary technology can exist in badly managed companies. Amazon survived the dot-com crash. Pets.com didn’t.
3. Ignoring Historical Patterns
“This time is different because AI is real technology.”
Reality: The internet was also real technology. The valuations were still fake. Bubbles form around real things.
4. Assuming Current Pricing Persists
“API costs will keep dropping.”
Reality: When profitability pressure arrives, costs often increase. Uber was cheap until they needed profit.
Summary
I started this research trying to answer one question: Is the AI investment bubble going to burst?
The answer is yes. The mechanism is clear: profitability pressure forces monetization, monetization triggers enshittification, enshittification causes churn, churn reveals revenue reality, reality causes valuations to collapse.
But that’s not the whole story.
The technology underneath the hype is genuine. Microsoft FrontPage produced crappy websites because it was a static template system. AI produces genuinely useful output because it learned from billions of examples. The comparison to previous “code generators” misses this fundamental difference.
For individuals: build transferable skills, stay diversified, ignore FOMO pressure, measure actual outcomes rather than promises.
For organizations: implement incrementally, calculate ROI, avoid lock-in, prepare for price increases.
The bubble will burst because that’s what bubbles do. The technology will survive because that’s what useful technology does. Both things. At the same time. That’s not a contradiction—it’s the history of every transformative technology that ever went through an investment bubble.
Final Words + More Resources
My intention with this article was to help others share my knowledge and experience. If you want to contact me, you can contact by email: Email me
Here are also the most important links from this article along with some further resources that will help you in this scope:
- 👨💻 Reddit Discussion: Truly Weird Tech Times
- 👨💻 Dot-com Bubble Historical Analysis
- 👨💻 Microsoft FrontPage History
- 👨💻 Enshittification Theory - Cory Doctorow
Oh, and if you found these resources useful, don’t forget to support me by starring the repo on GitHub!
Comments