⚡ Quick Summary
- AI founders are issuing equity at two different effective prices simultaneously — a small high-priced tranche sets the headline valuation, while a larger lower-priced tranche (via SAFEs, convertibles, or structured preferred shares) provides institutional investors with downside protection.
- Secondary market platforms including Forge Global and Carta are showing AI startup shares trading 30–60% below their most recent primary round prices, revealing the real-world gap between manufactured and fundamental valuations.
- Enterprise IT departments face operational risk when selecting AI vendors based on unicorn status as a proxy for financial stability — a manufactured $2 billion valuation offers no genuine protection against vendor collapse.
- The SEC and FCA are both examining private market valuation disclosure practices, and the dual-pricing mechanism sits in legally ambiguous territory that could attract regulatory enforcement as AI IPO filings begin to expose full capital structures.
- AI companies with genuine ARR traction and transparent capital structures — including players like Glean and Cohere — have a significant opportunity to differentiate themselves as the manufactured-valuation bubble faces increasing scrutiny.
What Happened
A quiet but consequential financial engineering trend is sweeping through the AI startup ecosystem: founders are increasingly issuing the same class of equity at two materially different prices — simultaneously — to manufacture the appearance of billion-dollar valuations without the underlying business fundamentals to justify them. The mechanism, while not entirely new to venture capital, has taken on an alarming new scale and sophistication in the AI boom of 2023–2025.
The core mechanics work roughly like this: a startup raises a small, highly priced round — often from a friendly investor, a strategic partner, or even a related party — at a valuation that establishes a headline number. That headline number, say $1.2 billion, becomes the company's public identity. A second, larger tranche of capital is then raised at a significantly lower effective price, frequently through structured instruments like SAFEs (Simple Agreements for Future Equity), convertible notes with aggressive discount rates, or preferred share classes carrying substantial liquidation preferences and anti-dilution ratchets. The investor in the lower-priced tranche is protected on the downside; the founder gets the unicorn badge. Everyone, at least temporarily, wins — except the uninformed observer who takes the headline valuation at face value.
What makes this particularly acute in the current AI landscape is the sheer volume of capital chasing a relatively small number of credible AI infrastructure and application layer companies. With sovereign wealth funds, corporate venture arms from Microsoft, Google, Amazon, and Salesforce, and traditional tier-one VCs all competing for allocation in AI deals, founders have unprecedented leverage to dictate terms. That leverage is now being used not just to secure better economics, but to engineer perception itself.
The practice came into sharper focus in early 2025 as several AI startups that had announced unicorn or decacorn status faced scrutiny from secondary market platforms — including Carta, Forge Global, and Nasdaq Private Market — where their shares were trading at significant discounts to the last reported primary round price, in some cases 40 to 60 percent lower.
Background and Context
To understand why this is happening now with such intensity, you have to trace the arc of AI investment from roughly 2022 onward. The release of OpenAI's ChatGPT in November 2022 functionally triggered a Cambrian explosion in AI startup formation. By Q1 2023, CB Insights was tracking over 1,000 new AI-focused startups raising seed rounds. By Q4 2024, PitchBook data indicated that AI companies accounted for approximately 35 percent of all US venture capital deployed — a staggering concentration in a single thematic category.
This wasn't the first time the startup ecosystem had seen valuation manipulation. The 2015–2019 unicorn bubble, driven largely by SoftBank's Vision Fund deploying $100 billion at aggressive valuations, produced a generation of companies — WeWork, Oyo, Greensill — whose paper valuations bore little relationship to their intrinsic worth. The Ratchet Clause era of 2016–2018 saw similar dual-pricing dynamics, where late-stage investors extracted IPO-price guarantees and liquidation preferences that effectively meant they owned a different, safer asset than common shareholders while both were nominally priced at the same per-share figure.
What's different in the current AI cycle is the speed and the technical opacity. AI companies can credibly claim that their models, datasets, and inference infrastructure are difficult to value using traditional metrics — ARR multiples, customer acquisition costs, churn rates. When a company says its large language model fine-tuning pipeline represents a defensible moat, it's genuinely hard for a generalist LP or a journalist to interrogate that claim. This technical complexity creates cover for financial engineering that would be more visible in, say, a SaaS company with straightforward subscription metrics.
The SAFE instrument, originally designed by Y Combinator in 2013 to simplify early-stage fundraising, has become a particular vehicle of concern. SAFEs issued with valuation caps of $50 million that convert into a round priced at $500 million represent a 10x dilution event that doesn't show up cleanly in the headline valuation story. Founders and their communications teams announce the $500 million round; the SAFE holders quietly receive their shares at a fraction of that price.
Why This Matters
The implications of this trend extend well beyond the venture capital community, touching enterprise technology buyers, corporate IT departments, and the broader technology industry in ways that aren't immediately obvious.
First, and most directly, enterprise procurement decisions are increasingly influenced by a vendor's perceived financial stability and market credibility. When a Fortune 500 company's IT department is evaluating an AI platform for integration into its Microsoft 365 environment — perhaps for Copilot augmentation, custom RAG pipelines, or enterprise search — the vendor's unicorn status functions as a proxy for durability. A $2 billion valuation implies institutional backing, runway, and the ability to support enterprise SLAs. If that valuation is manufactured through dual-pricing mechanics, the enterprise customer has made a vendor selection based on false signals. The risk of vendor lock-in to a company that subsequently collapses or undergoes a distressed acquisition is very real.
Second, this dynamic distorts the competitive landscape for legitimate AI companies with genuine traction. Startups that have earned their valuations through real ARR growth — companies like Glean, which reportedly crossed $100 million ARR in 2024, or Cohere, which has built meaningful enterprise contracts — are competing for talent, customers, and press attention against companies whose billion-dollar headlines are partially illusory. The signal-to-noise ratio in the AI vendor market degrades, making it harder for enterprise buyers to identify genuinely sound investments of their procurement dollars.
Third, for IT professionals managing technology stacks that increasingly incorporate AI components — whether through Microsoft's Azure OpenAI Service, Google's Vertex AI, or third-party AI middleware — the instability risk is operational, not just financial. An AI vendor that implodes mid-contract can leave organisations with broken integrations, stranded data, and compliance exposure. Businesses managing enterprise productivity software stacks should be conducting deeper financial due diligence on AI vendors than the headline valuation alone suggests.
The regulatory dimension is also sharpening. The SEC has been increasingly attentive to private market valuation practices since its 2023 guidance on private fund advisers, and the UK's FCA has signalled similar interest in how private valuations are communicated to institutional investors. The dual-pricing mechanism sits in a legally ambiguous space — it isn't necessarily fraudulent, but when the lower-priced instruments are not prominently disclosed in investor communications, it approaches the territory of material omission.
Industry Impact and Competitive Landscape
The ripple effects of this valuation engineering are being felt across the entire AI investment and deployment ecosystem, and the major technology platforms are not immune to the distortions it creates.
Microsoft, which has committed approximately $13 billion to OpenAI across multiple tranches since 2019, has a direct interest in the credibility of AI valuations. OpenAI's own valuation trajectory — from $29 billion in early 2023 to $157 billion in the October 2024 funding round led by Thrive Capital — has itself attracted scrutiny, particularly given that the round included provisions reportedly allowing investors to sell back shares if OpenAI did not convert to a for-profit structure within two years. This is precisely the kind of structured instrument that creates a two-tier ownership reality beneath a single headline number.
Google DeepMind, Anthropic (in which Google has invested approximately $2 billion and Amazon a further $4 billion), and Meta's internal AI research organisation all compete in an environment where the perceived scale of competitor funding shapes talent recruitment, partnership negotiations, and customer confidence. Inflated valuations at competitor startups force these organisations to respond — either by inflating their own investment announcements or by accelerating product releases to demonstrate tangible value against paper unicorns.
The secondary market platforms — Carta, Forge Global, and EquityZen — are caught in an uncomfortable position. They facilitate liquidity for startup employees and early investors, and their pricing data increasingly contradicts the primary market valuations that founders announce. In several documented cases in 2024, AI startup shares trading on Forge were clearing at 30 to 50 percent below the most recent primary round price within six months of that round closing. This secondary market discount is the market's honest assessment of what the structured instruments are worth when you strip away the liquidation preferences and ratchets.
For Salesforce, which has been aggressively building its own AI portfolio through Salesforce Ventures and the Einstein AI platform, the valuation distortions create M&A complexity. Acquiring an AI startup that has manufactured unicorn status means paying a premium that doesn't reflect the actual equity waterfall — the structured investors at the lower price will be made whole first in any acquisition, leaving less for common shareholders and potentially making the deal economics unattractive at the headline price.
Expert Perspective
From a structural finance standpoint, what's occurring in the AI startup market is a sophisticated form of information asymmetry exploitation. Founders and their lead investors possess complete knowledge of the capital stack — every SAFE, every convertible note, every liquidation preference multiplier. The market, including downstream enterprise customers, journalists, and prospective employees, sees only the headline valuation.
Industry analysts at firms like Gartner and Forrester have begun incorporating "valuation quality" assessments into their vendor evaluations, recognising that the traditional proxy of funding round size is no longer a reliable signal. Gartner's 2024 Hype Cycle for Artificial Intelligence explicitly flagged the gap between AI startup valuations and demonstrable enterprise value delivery as a key risk factor for technology buyers.
The deeper strategic risk is what happens when the AI investment cycle turns, as cycles inevitably do. The 2022 correction in public technology markets — when the Nasdaq Composite fell approximately 33 percent and cloud software multiples compressed from 20x ARR to 6x ARR in under twelve months — demonstrated how rapidly sentiment can shift. When that correction arrives in private AI markets, companies that have manufactured valuations through structured instruments will face a particularly brutal reckoning: the structured investors are protected, common shareholders are wiped out, and the enterprise customers who bet on their stability are left holding broken integrations and voided contracts.
The opportunity, paradoxically, is for AI companies with genuine fundamentals to differentiate themselves through radical transparency — publishing ARR, net revenue retention, and customer count data that their manufactured-valuation competitors cannot credibly match.
What This Means for Businesses
For enterprise technology decision-makers, the practical response to this environment requires updating vendor due diligence frameworks that were designed for a simpler era. Asking an AI vendor for their last funding round amount is no longer sufficient. The right questions are: What is the full capital structure, including all SAFEs and convertible instruments? What are the liquidation preferences on your preferred stock? What is your current ARR and net revenue retention rate? Who are your five largest enterprise customers, and can we speak with them?
IT departments evaluating AI tools for integration with Microsoft 365 environments, Azure infrastructure, or Windows-based workflows should treat vendor financial stability as a first-class evaluation criterion alongside technical capability. An AI middleware provider that scores brilliantly on benchmark evaluations but has a precarious capital structure represents an operational risk that technical merit alone cannot offset.
Businesses should also consider the total cost of ownership implications of AI vendor instability. Migration costs, retraining costs, and integration rebuild costs can easily exceed the initial licensing savings from choosing a cheaper, less stable vendor. Where budget is a constraint — and it always is — there are legitimate ways to optimise spending on foundational software. For instance, securing an affordable Microsoft Office licence through a reputable reseller can free up budget for more rigorous AI vendor evaluation and procurement processes, rather than defaulting to whichever AI vendor has the most impressive headline valuation.
The broader lesson is that in a market flooded with manufactured signals, genuine due diligence has never been more valuable — or more differentiating.
Key Takeaways
- Dual-pricing mechanics are structural, not accidental: AI founders are deliberately using SAFEs, convertible notes, and tiered preferred shares to establish high headline valuations while giving institutional investors downside protection at lower effective prices.
- Secondary market data is the honest signal: Platforms like Forge Global and Carta are showing AI startup shares trading 30–60% below primary round prices, revealing the gap between manufactured and real valuations.
- Enterprise procurement is directly affected: Vendor selection decisions based on unicorn status as a stability proxy may be systematically flawed, creating operational risk for IT departments building AI integrations.
- Regulatory scrutiny is intensifying: Both the SEC and FCA are examining private market valuation disclosure practices, and the dual-pricing mechanism sits in legally ambiguous territory that could attract enforcement action.
- The AI investment cycle correction will be uneven: When sentiment shifts, structured investors will be protected while common shareholders and enterprise customers bear disproportionate risk.
- Transparency is a competitive moat: AI companies with genuine ARR traction and clean capital structures have an opportunity to differentiate by publishing financial metrics their manufactured-valuation competitors cannot credibly match.
- Due diligence frameworks need updating: Asking only about funding round size is obsolete; enterprise buyers must interrogate the full capital stack, liquidation preferences, and actual revenue metrics.
Looking Ahead
Several developments in the next six to eighteen months will determine whether this dual-pricing trend accelerates, stabilises, or triggers a market correction. The most significant near-term catalyst is the anticipated wave of AI startup IPOs expected in 2025–2026. When these companies file S-1 registration statements with the SEC, the full capital structure — every SAFE, every liquidation preference, every ratchet clause — becomes public record. The gap between the disclosed structure and the previously announced headline valuations will be visible to anyone who reads the filing. For some companies, that disclosure will be unremarkable; for others, it will be damaging.
Watch also for the evolution of Microsoft's Azure OpenAI Service pricing and capability roadmap, which is scheduled for significant updates through 2025. As Microsoft continues to embed AI capabilities directly into Windows 11 and Microsoft 365 — including the Copilot+ PC initiative and the expanding Copilot for Microsoft 365 feature set — the value proposition of standalone AI startups will face increasing pressure from platform-native alternatives. Ensuring your organisation has a genuine Windows 11 key and is running current Microsoft infrastructure will be foundational to accessing these integrated AI capabilities as they roll out.
The secondary market data will be the leading indicator. Track the Forge Global and Carta pricing indices for AI sector shares through Q3 2025 — if the discount to primary round prices widens beyond 50 percent on a sustained basis, a broader reckoning in AI startup valuations is likely approaching.
Frequently Asked Questions
What exactly is the dual-pricing equity mechanism that AI startups are using?
The mechanism involves raising capital in two effectively separate tranches at different prices, even though both are nominally part of the same company's equity. A small round is priced at a high valuation to establish a headline number — say $1 billion — while a larger amount of capital is raised through instruments like SAFEs (Simple Agreements for Future Equity), convertible notes with aggressive discount rates, or preferred share classes with heavy liquidation preferences. The second group of investors receives shares at a much lower effective price per unit of ownership, giving them downside protection. The founder announces the high headline valuation publicly; the structured nature of the second tranche is typically disclosed only in legal documents that receive little public attention.
Why is this particularly prevalent in AI startups right now?
Several factors converge to make AI startups especially susceptible to this practice. First, AI companies can credibly claim their technology — large language models, proprietary datasets, inference infrastructure — is difficult to value using traditional metrics like ARR multiples or customer acquisition costs, creating cover for financial engineering. Second, the sheer volume of capital chasing a limited number of credible AI deals gives founders unusual leverage to dictate terms. Third, the speed of the current AI investment cycle — driven by the post-ChatGPT boom since late 2022 — has compressed the due diligence timelines that would normally surface these structures. Finally, unicorn status in AI carries outsized reputational value for talent recruitment, enterprise sales, and media coverage, creating strong incentives to manufacture it.
How should enterprise IT departments protect themselves when evaluating AI vendors?
Enterprise buyers should update their vendor due diligence frameworks to go well beyond headline valuation figures. Key questions to ask any AI vendor include: What is the complete capital structure including all SAFEs and convertible instruments? What are the liquidation preference multiples on preferred stock? What is the current ARR and net revenue retention rate? Who are the five largest enterprise customers and can references be provided? IT departments should also monitor secondary market trading data on platforms like Forge Global for any AI vendor whose shares are publicly traded there — sustained discounts to primary round prices are a leading indicator of valuation stress. Treating vendor financial stability as a first-class evaluation criterion alongside technical capability is essential when building AI integrations into critical business infrastructure.
What will force this issue into the open, and when?
The most significant forcing function will be AI startup IPO filings, which are expected to accelerate through 2025 and 2026. When companies file S-1 registration statements with the SEC, the complete capital structure — every SAFE, every liquidation preference, every anti-dilution ratchet — becomes public record. For some companies, the disclosed structure will match the narrative; for others, the gap will be significant and damaging. Regulatory developments are also a factor: the SEC's 2023 guidance on private fund advisers and the FCA's ongoing review of private market valuation practices are both moving toward greater disclosure requirements. Secondary market pricing data on platforms like Forge Global and Carta will serve as the leading indicator — watch for AI sector discounts to primary round prices widening beyond 50% on a sustained basis as an early warning signal.