‘Vibe revenue’: AI companies admit they’re worried about a bubble
‘Vibe revenue’: AI companies admit they’re worried about a bubble
The chief executive officers of artificial intelligence firms DeepL and Picsart have publicly expressed concerns about a potential valuation bubble in the AI sector. In statements to CNBC, they acknowledged the high valuations and market hype while also affirming their belief in the long-term potential of the technology. The term 'vibe revenue' has emerged to describe valuations driven by narrative and future promise rather than current financial performance.
Context & What Changed
The period following the public release of advanced generative AI models in late 2022, such as OpenAI’s ChatGPT, initiated an unprecedented investment surge into the artificial intelligence sector. This boom has been characterized by rapidly escalating corporate valuations, intense competition for specialized talent, and significant capital expenditure on computational infrastructure. Global private investment in generative AI alone reached $25.2 billion in 2023, a nearly nine-fold increase from 2022 (source: hai.stanford.edu). This influx of capital has pushed valuations for AI startups to levels often disconnected from traditional financial metrics like revenue or profitability, leading to the coining of terms like ‘vibe revenue’—where a company’s perceived value is based on market excitement, narrative strength, and speculative future potential rather than demonstrated business fundamentals.
The significant change highlighted by the source article is the shift of this bubble narrative from the domain of external market analysts and skeptics to that of industry insiders. When the CEOs of prominent, well-funded AI companies like DeepL (a language technology company) and Picsart (a digital creation platform) publicly acknowledge concerns about a potential bubble, it signals a new phase of market maturity and self-awareness. This internal validation of risk lends significant credibility to concerns about market froth and suggests that industry leaders are beginning to plan for a potential market correction or rationalization. This is no longer just external critique; it is an internal acknowledgement of unsustainable trends.
Stakeholders
1. Governments & Regulators: National and international bodies (e.g., U.S. SEC, UK’s FCA, European Commission) are concerned with financial stability, investor protection, and the macroeconomic impact of a potential tech bubble burst. A disorderly correction could erase significant paper wealth, impact employment in key tech hubs, and potentially require regulatory or monetary policy responses. They are also tasked with fostering genuine innovation while preventing market manipulation or systemic risk.
2. Large-Cap Technology Actors: Incumbents like Microsoft, Alphabet (Google), Amazon, and Meta are primary drivers and beneficiaries of the AI boom through direct investment, partnerships (e.g., Microsoft’s investment in OpenAI), and providing essential cloud infrastructure. They are heavily exposed to a downturn, both through their direct investments and the potential for reduced demand for their high-margin cloud and software services if AI startups fail.
3. Semiconductor & Hardware Providers: Companies like Nvidia have seen their market capitalization soar due to the extreme demand for specialized GPUs required for training and running AI models. A slowdown in AI development or a shift towards more efficient models would directly impact their revenue forecasts and stock valuations.
4. AI Startups & Scale-ups: These are the direct subjects of the high valuations. While benefiting from access to capital, they face immense pressure to grow into their valuations. A market correction would severely restrict their access to funding, forcing layoffs, pivots, or outright failure for those with high cash-burn rates and no clear path to profitability.
5. Venture Capital (VC) & Private Equity (PE) Firms: These investors have allocated a substantial portion of their funds to AI. A bubble burst would lead to significant portfolio write-downs, impacting their limited partners (pensions, endowments) and making future fundraising more difficult. The entire VC business model relies on high-valuation exits, which would become scarce in a downturn.
6. Infrastructure Providers: Data center operators (e.g., Digital Realty, Equinix) and energy utilities are planning massive capital expenditures based on forecasts of exponential growth in AI-driven computational demand. A correction would lead to project cancellations or delays, potentially resulting in stranded assets and revenue shortfalls.
7. Enterprise Adopters: Non-tech corporations investing in AI solutions to improve productivity face the risk of partnering with or building on platforms of vendors who may not be viable long-term. A vendor’s failure could disrupt critical business operations and waste significant investment.
Evidence & Data
The thesis of an AI bubble is supported by several quantitative and qualitative data points:
– Capital Inflows: Total global private investment in AI in 2023 was $91.9 billion (source: CB Insights). While down from a 2021 peak, the concentration in the generative AI sub-sector shows intense, focused speculation.
– Valuation Metrics: AI startups have commanded valuation multiples far exceeding historical norms for software companies. For instance, some generative AI firms have been valued at over 100 times their annual recurring revenue (ARR), compared to a typical 10-25x for a high-growth public SaaS company. OpenAI’s reported valuation of over $80 billion in early 2024 against reported revenues of around $2 billion for 2023 exemplifies this trend (source: various media reports, e.g., Reuters, Bloomberg).
– Concentration Risk: The market’s gains have been highly concentrated. The stock of Nvidia, which produces an estimated 80% of the GPUs used for AI training, increased by approximately 240% in 2023 alone, indicating heavy market reliance on a single point of the value chain (source: market data providers).
– High Costs & Unproven Business Models: The cost to train a frontier AI model can exceed $100 million in compute resources alone (source: Epoch AI research). This necessitates continuous, large-scale funding rounds, often before a sustainable, profitable business model has been identified. Many generative AI applications currently struggle with unit economics, where the cost of inference (running the model for a user) can exceed the revenue generated per query.
– Insider Commentary: The primary evidence from the source article—the public statements of concern from DeepL CEO Jaroslaw Kutylowski and Picsart CEO Hovhannes Avoyan—is a crucial qualitative indicator. Such admissions are rare during a boom and suggest that the pressures and irrationality of the current environment are becoming internally acknowledged.
Scenarios (3) with probabilities
1. Scenario 1: Soft Landing / Rationalization (Probability: 50%)
In this scenario, the market undergoes a correction, not a crash. A flight to quality occurs, where investment shifts from speculative ‘story’ stocks to AI companies with demonstrable product-market fit, strong revenue growth, and a clear path to profitability. Valuations for second- and third-tier players decline, leading to down-rounds and consolidation through acquisitions by larger tech firms. The overall pace of AI innovation continues, but with a newfound emphasis on capital efficiency and sustainable business models. The impact on public markets is contained within the tech sector, and there is no systemic financial contagion.
2. Scenario 2: Contained Bubble Burst (Probability: 35%)
This scenario involves a significant and rapid price collapse, primarily impacting privately-held AI startups and recently-public AI-focused companies. Widespread startup failures occur, leading to substantial losses for VC and PE funds. This triggers a broader tech sector downturn, reminiscent of the dot-com bust of 2000-2001, but with less systemic impact due to the stronger balance sheets of today's tech giants (e.g., Apple, Microsoft). Mass layoffs in the tech sector occur, impacting regional economies like the San Francisco Bay Area. Infrastructure projects tied to AI are postponed. The recovery is slower, with a 2-3 year period of reduced investment and innovation.
3. Scenario 3: Systemic Contagion (Probability: 15%)
The AI bubble bursts violently, and its effects spill over into the broader financial system. The immense market capitalization of AI-adjacent firms like Nvidia is revealed to be a key support for the entire stock market. A sharp decline in these stocks triggers a wider market crash. The interconnectedness of large banks, institutional investors, and the tech sector creates a domino effect. This scenario could be exacerbated if the AI boom has masked underlying weaknesses in the global economy. Central banks may be forced to intervene to provide liquidity. This would have severe consequences for public finance, infrastructure delivery, and global economic growth.
Timelines
– Short-Term (0-12 months): Expect increased investor scrutiny on AI company metrics beyond user growth, focusing on revenue, gross margins, and customer retention. We may see the first high-profile down-rounds or fire sales of struggling startups. Regulators like the SEC will likely increase warnings to retail investors about the speculative nature of some AI investments.
– Medium-Term (1-3 years): This is the most probable window for a significant market correction or burst (Scenarios 2 or 3). Many AI startups funded in the 2022-2023 boom will exhaust their cash reserves and need to demonstrate commercial viability to secure further funding. Consolidation will accelerate. The true, durable AI platforms will begin to separate from the hype.
– Long-Term (3+ years): The market will have stabilized. Surviving companies will emerge as new industry leaders with proven business models. The genuine productivity gains from AI will be more clearly measurable and integrated into the economy. Mature regulatory frameworks governing AI development and deployment will be in place.
Quantified Ranges (if supported)
– Private Investment Correction: In a ‘Contained Burst’ scenario, annual private AI investment could fall by 40-60% from its peak, returning to pre-2023 levels. This implies a drop from over $90 billion to a range of $35-55 billion annually.
– Valuation Multiples: Median revenue multiples for AI startups could compress significantly, from peaks of 50-100x ARR to a more sustainable 8-20x ARR range, in line with historical SaaS market norms.
– Data Center Investment: A slowdown could defer or cancel $100 billion to $250 billion in planned global data center construction over the next five years (author’s estimate based on combining public growth projections from major operators with correction scenarios).
Risks & Mitigations
– Risk: Capital Misallocation & Innovation Stagnation: Billions are invested in redundant or unsustainable AI models, diverting resources from more fundamental research or viable applications.
– Mitigation: Investors must enforce stricter due diligence, focusing on proprietary technology, data moats, and unit economics. Governments can support basic research and open standards to prevent a winner-take-all dynamic based on capital alone.
– Risk: Financial Instability: A sudden collapse in AI-related asset prices triggers a broader market panic and credit crunch.
– Mitigation: Financial regulators should conduct stress tests on institutional exposure to the high-valuation tech sector. Corporate boards of non-tech firms should ensure their balance sheets are not overly exposed to the volatility of tech stocks used for treasury management.
– Risk: Infrastructure Overbuild & Stranded Assets: Energy and data center capacity is built based on hype-driven demand forecasts that fail to materialize.
– Mitigation: Infrastructure providers should pursue modular designs and phased build-outs tied to concrete, long-term customer commitments. Policymakers must integrate data center energy needs into long-term grid planning to ensure stability and avoid subsidizing speculative builds.
Sector/Region Impacts
– Sectors: The Technology sector, particularly Software and Semiconductors, would be most directly and severely impacted. Financial Services, especially VC firms, would face a cyclical downturn. Energy and Utilities would see a slowdown in a key demand growth driver. Conversely, a correction could benefit Enterprise Adopters in the long run by weeding out weaker vendors and lowering the cost of proven AI solutions.
– Regions: North America, specifically the US West Coast, as the global epicenter of AI funding and development, would bear the brunt of a downturn in terms of job losses and economic impact. East Asia (Taiwan, South Korea) would be affected via the semiconductor supply chain. Europe’s focus on regulation (e.g., EU AI Act) could either prove to be a stabilizing force that encouraged more sustainable AI development or be blamed for a lack of competitiveness, depending on the outcome.
Recommendations & Outlook
For Policymakers & Regulators:
– Enhance investor education regarding the risks of speculative technology markets. Avoid direct market intervention but ensure transparency in financial reporting for AI companies.
– (Scenario-based assumption) Assuming a ‘Soft Landing’ is the most probable outcome, policy should focus on the long-term enablers of AI: STEM education, public research funding, and developing standards for AI safety and interoperability.
For Infrastructure Providers (Energy, Data Centers):
– (Scenario-based assumption) Given the significant risk of a ‘Contained Burst,’ investment models must be de-risked. Prioritize clients with strong balance sheets and proven business cases. Link infrastructure expansion directly to secured, long-term contracts rather than speculative market growth.
For Corporate Boards & CFOs:
– Adopt a disciplined, ROI-centric approach to all AI investments. Distinguish between exploratory R&D and enterprise-scale deployment, demanding clear business cases for the latter.
– (Scenario-based assumption) To prepare for market rationalization, conduct due diligence on the financial viability of key AI vendors and partners. Develop contingency plans for a scenario where a critical AI provider is acquired or fails.
Outlook:
The long-term potential of artificial intelligence to drive productivity and economic growth remains exceptionally high. However, the current investment cycle is exhibiting classic signs of speculative excess. The acknowledgement of bubble risks by industry leaders is a pivotal moment, likely heralding a forthcoming period of rationalization. The central strategic challenge for all stakeholders is to navigate this correction—to separate the durable, value-creating aspects of AI from the transient market hype. Organizations that maintain financial discipline, focus on solving real-world problems, and manage their exposure to systemic risks will be best positioned to thrive in the more mature AI landscape that will emerge.