📊 Full opportunity report: The Bubble Question, Disentangled: 1999 vs 2026 Category by Category on ThorstenMeyerAI.com — validation score, market gap, and execution plan.
TL;DR
This analysis compares the AI investment landscape of 2026 with the dotcom bubble of 1999, highlighting where bubble risks exist and where genuine value is emerging. It emphasizes the importance of category-specific assessments for future strategies.
Recent assessments indicate that the 2026 AI investment cycle exhibits both bubble-like and fundamentally grounded characteristics, echoing but also diverging from the 1999 dotcom bubble. Experts emphasize that disentangling these categories is critical for understanding risks and opportunities for investors, policymakers, and industry leaders.
Key indicators such as private valuations, capital deployment, and market concentration reveal that certain segments of AI investment, notably private startups and infrastructure buildout, show bubble-like traits similar to the late 1990s. For instance, private valuations for AI firms have reached hundreds of billions of dollars, with mega-deals dominating venture capital funding—paralleling the dotcom era’s speculative frenzy.
However, unlike 1999, the current cycle benefits from tangible revenue, visible productivity gains, and real earnings growth from established players like the Magnificent Seven. Multiple expansion plays a smaller role, and fundamentals such as revenue and cash flow are more grounded, suggesting some sectors are less prone to collapse.
Experts like Thorsten Meyer highlight that the cycle is bifurcated: some categories, such as infrastructure and private valuations, are in bubble territory, while others, like enterprise AI deployment and productivity, reflect durable value. This nuanced view challenges the binary ‘bubble or not’ narrative, emphasizing category-specific analysis.
Not binary.
Category by category.
Some bets show clear bubble dynamics. Some show durable value. The disentanglement matters more than the aggregate framing.
OpenAI $730B private valuation. Anthropic $380B. Mag 7 forward P/E 38× vs Dot-com peak 30×. BUT: earnings-driven returns (78%) vs Dot-com multiple-driven (314%). Real productivity gains. Mag 7 outsized free cash flow. Carlota Perez framing applies.
Two cycles. Twelve dimensions.
On price-and-fundamentals dimensions, 2024-2026 is more grounded than 1999. On capital-allocation dimensions, 2024-2026 has bubble-comparable or worse characteristics. The dual signal explains the analyst disagreement.
Your AI Survival Guide: Scraped Knees, Bruised Elbows, and Lessons Learned from Real-World AI Deployments
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Five frothy. Five durable. Three contested.
The honest read: the cycle is structurally bifurcated. Some categories are not in bubble territory; others are. The contested middle is where the bubble question actually resolves through 2027-2028.
- Mega-deal concentrationOpenAI $730B, Anthropic $380B, Databricks $134B.
- Circular financingMSFT→OpenAI→CoreWeave→NVDA→MSFT loop.
- Capex velocity$725B exceeds revenue translation. $1.5T debt by 2028.
- Cahn / Sequoia argument$5T buildout requires AGI by 2030.
- Capital-flow speed$700B retail equity since Jan · 5× faster than 2000.
- Hyperscaler capex justificationCahn (only AGI) vs Goldman (justified by trajectory).
- NVIDIA addressable shareCUDA moat vs in-house silicon migration to 30-45% by 2028.
- Frontier-lab valuationsPlatform companies vs commodity API providers.
- Earnings-driven returns78% earnings · 9% multiples vs Dot-com 314% multiples.
- Mag 7 FCF + buybacksMicrosoft $90B FCF · Alphabet $70B · structural cushion.
- Profit weight matchesTech ~30% market cap, ~20% profits vs 1999 35%/10% gap.
- Forward margins recordS&P Tech margin estimates at all-time highs.
- Real productivity30-50% call center · 20-40% software eng · measurable today.
Opengear OM2248-10G Infrastructure Management Equipment
Autosensing is very impressive
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Three paths. One question.
35/50/15 probability. Base scenario most likely because durable-value supports prevent worst-case but bubble signals are too strong to resolve without correction.
- Frothy correct 30-50%Frontier labs, circular financing.
- Mag 7 sustainsReal productivity continues.
- Hyperscaler capex defensibleMixed but justified.
- NVIDIA gradual decelNot sharp.
- Outcome: Uneven returns. Big winners + losers. No broad crash.
- Frontier labs -40-60%From 2026 peaks.
- Hyperscaler impair$50-150B capex aggregate.
- NVIDIA sharp decelFY28 30-50% growth vs FY26 75%.
- NASDAQ -30-50%12-24 month period.
- Outcome: Mag 7 cushion holds. Deployment continues delayed.
- NASDAQ -60-78%Matching 2001-2003 magnitude.
- Frontier labs collapseBelow VC entry pricing.
- Hyperscaler impair $300-500BMajor capex writedowns.
- NVIDIA negative quartersRevenue compression.
- Outcome: Multi-year recovery. Deployment 2032-2033.
The 2024-2026 cycle is structurally more grounded than 1999 on price-and-fundamentals dimensions and structurally similar or worse on capital-allocation dimensions. The bifurcation explains the analyst disagreement and predicts the correction pattern: specific categories correct sharply while others persist.
private AI startup valuation reports
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Four assignments. By role.
Stop pricing AI as single asset class.
Differentiate Mag 7 (durable-value-leaning) from pure-play AI infrastructure (bubble-leaning) from contested middle (NVIDIA, frontier labs). Position long durable-value categories; short or underweight bubble-categories with circular-financing exposure. Use Perez framing to size correction expectations.
Pace through 2026-2027.
Preserve dry powder for 2028-2029. Mega-rounds at $300B+ valuations carry asymmetric correction risk. Mid-stage product-market-fit names with real revenue carry durable value through any plausible correction. The 1999 lesson: winners eventually recover; losers don’t.
Build for survivable correction.
18-24 month cash runway assumptions that survive 30-50% valuation correction. Prioritize real revenue over narrative-driven funding. Structure cap tables to absorb down-round scenarios. Peak-fundraising window of 2025-2026 may not persist; raise opportunistically while it does.
Multi-vendor sourcing for price volatility.
Plan for AI service price volatility through 2027-2028. Prices may rise (power constraint) or fall (frontier-lab competitive pressure). Multi-vendor sourcing reduces single-vendor exposure. Contractual flexibility (escalators, exit provisions, renegotiation triggers) preserves optionality.
AI productivity software
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Implications of Category-Specific Bubble Risks
This nuanced assessment matters because it guides strategic decisions across sectors. Investors need to differentiate between overhyped startups and infrastructure investments likely to persist. Policymakers must understand where regulation or intervention may be needed to prevent systemic risks. For industry leaders, recognizing which AI segments offer durable value versus speculative excess influences R&D, deployment, and investment priorities.
Historical and Current Market Dynamics of AI Investment
The 1999 dotcom bubble was characterized by excessive venture capital deployment—$54 billion in 1999—with 62% flowing into unprofitable firms and a surge of IPOs at valuations detached from fundamentals. When the bubble burst, many firms, including Pets.com and Webvan, collapsed, but durable companies like Amazon and Cisco eventually recovered and thrived.
In contrast, the 2024-2026 AI cycle features record private valuations, with OpenAI valued at $730 billion and Anthropic at $380 billion, alongside extreme concentration of VC funding—73% of AI venture capital in a handful of firms. Infrastructure spending has surged to $725 billion in 2026, comparable in scale but faster-paced than 1999.
Unlike the dotcom era, where multiple expansion was predominant, the current cycle shows more earnings-driven growth, though capital allocation remains risky with high concentration and circular financing patterns. This divergence underscores the importance of granular analysis to avoid conflating bubble risks with genuine technological progress.
“The cycle is structurally bifurcated; some categories are bubble-prone, others reflect real value. Disentangling these is key to navigating the next few years.”
— Thorsten Meyer
Unclear Aspects of the 2026 AI Cycle
While some AI segments show clear bubble characteristics, it remains uncertain how many current valuations are justified by future earnings and productivity gains. The timing and scale of potential corrections are still developing, especially given the rapid infrastructure investments and private valuations that are difficult to evaluate against traditional metrics. Additionally, the impact of potential regulation or technological breakthroughs on these valuations is still unknown.
Future Developments and Monitoring Indicators
Investors and policymakers should closely monitor valuation trends, capital deployment patterns, and infrastructure spending over the next 12-24 months. Key indicators include private valuation adjustments, VC exit activity, and the evolution of revenue and earnings from core AI applications. The next major milestones will include earnings reports from leading AI firms, updates on infrastructure projects, and regulatory developments that could influence market dynamics.
Key Questions
How do current AI valuations compare to the dotcom bubble?
Private valuations are significantly higher, with firms like OpenAI valued at hundreds of billions, and VC funding is highly concentrated, resembling bubble traits from 1999. However, current fundamentals such as revenue and productivity gains are stronger than in the dotcom era.
Which AI sectors are most at risk of a bubble burst?
Private startups with extreme valuations, infrastructure buildouts, and sectors with high concentration of VC funding are most vulnerable to corrections if fundamentals fail to meet expectations.
What are the signs that the bubble might be deflating?
Signs include valuation adjustments, reduced VC funding, and a slowdown in infrastructure spending without corresponding revenue growth. Regulatory interventions could also accelerate corrections.
Is the current AI cycle sustainable?
Partially. While some segments show genuine value and productivity gains, others remain speculative. The sustainability depends on whether fundamentals can support valuations and whether technological breakthroughs materialize as expected.
What should investors focus on to avoid bubble risks?
Investors should emphasize fundamentals such as revenue, cash flow, and real productivity gains, and remain cautious of valuations detached from these metrics, especially in private markets and infrastructure investments.
Source: ThorstenMeyerAI.com
