Whether AI is a bubble has become a live, public argument among the venture investors funding it. In a recent StrictlyVC discussion published by TechCrunch, partners debated whether today's AI valuations are justified and whether ARR — the revenue metric those valuations lean on — is being inflated, per TechCrunch. The headlines treat this as a market-timing question for investors.
For anyone holding shares in a private AI-adjacent company, it's a more personal question. The valuation that defines your paper wealth — the number on your last 409A, the figure implied by your company's last round — is built on the same metrics now under dispute. This piece explains what the debate is actually about, what "ARR inflation" means in plain English, and why a fight over a revenue metric reaches all the way down to your own decision about your shares.
This is not tax advice, and it is not a recommendation to buy, sell, or hold any security. It is an explanation of how a valuation debate connects to the realizable value of private shares.
Key concept
A headline valuation is a mark, not a price you have realized. When that mark is justified largely by ARR — and the market is openly debating whether AI ARR is inflated — the durability of your paper wealth is exactly what's in question. The debate doesn't prove the number is wrong. It establishes that the number is uncertain. That distinction is the entire point for a shareholder: an uncertain mark is a reason to understand what your shares would actually clear at, not a reason to assume the headline holds.
What the debate is actually about
Start with what is publicly known, separate from interpretation. Investors are split in public. Some argue the technology is already delivering enough value to support its prices; others warn that, at the early stage especially, an AI narrative has been substituting for real traction and pulling valuations ahead of fundamentals. That disagreement — not a verdict — is the news.
Underneath the valuation argument sits a more specific one about a single metric: annual recurring revenue (ARR). ARR is supposed to capture the predictable, recurring subscription revenue a software company can count on. In the AI cohort, a separate TechCrunch report describes how founders and investors increasingly report padded ARR figures to crown category winners — and how some operators on the inside know it. One AI CEO put it bluntly: "To everyone who's inside, it just feels fake," per TechCrunch.
The two arguments are linked. If valuations are set as a multiple of ARR, and ARR itself is inflated, then the valuation inherits the inflation twice over — once in the metric, once in the multiple applied to it. That is why "is AI a bubble?" and "is AI ARR real?" are really the same question asked from two directions.
What "ARR inflation" actually means
ARR sounds precise. The practices that pad it are specific and, individually, often defensible — which is what makes the aggregate number hard to read. Here are the main ones, in plain English:
1. Counting CARR as ARR. CARR — contracted or committed ARR — includes deals that are signed but not yet live. Counting it as if it were real recurring revenue front-runs the risk that a customer cancels during a months-long implementation. One investor cited by TechCrunch observed CARR running 70% above actual ARR, with much of the gap never materializing.
2. Annualizing a single strong period. ARR is often built by taking one month — or one quarter, week, or even a single day — and multiplying it out to a full year. When demand is genuinely recurring, that's reasonable. When a customer can leave on 30 days' notice, multiplying one noisy month by twelve is closer to a hope than a measurement.
3. Treating pilots and trials as revenue. Free or heavily discounted pilots are how most enterprise AI gets adopted. Booking those pilots as paid recurring revenue counts dollars that may never recur once the pilot ends and a real contract is — or isn't — signed.
4. Annualizing usage-based revenue. Many AI products charge by usage rather than by seat or subscription. Usage revenue can swing 30–40% month to month, so annualizing the current month produces a figure that looks like a contract but behaves like a spot price.
5. Not netting out churn and downsell. An honest ARR figure subtracts expected customer losses and downgrades. A flattering one doesn't — for example, by counting the full value of a three-year deal that's deeply discounted in years one and two, on the assumption the customer stays to pay full price in year three.
Example
Consider a company reporting "$100 million ARR." On inspection, that figure might be $65 million of low-margin pass-through that vanishes if one large customer leaves, $25 million of usage revenue that swings month to month, and $10 million of genuinely contracted subscriptions. All three are real dollars today. Only one of them is the durable, recurring revenue the label implies — and a valuation set as a multiple of the full $100 million is pricing all three as if they were the last one.
None of this means every AI company is misrepresenting its numbers, and a defensible CARR or usage figure can be a perfectly honest way to describe a fast-growing business. The point is narrower: "ARR" is no longer a clean, comparable number across this cohort, so a valuation pegged to it carries more uncertainty than the single tidy figure suggests.
Why a metric debate reaches your cap table
Here is the connection that matters for a shareholder. The valuation attached to your company is not a price anyone has paid you. It is a mark — an estimate derived from the most recent priced round and the metrics used to justify it. You carry that mark in your head as your net worth. But you only ever realize a price when an actual buyer purchases your actual shares.
When the metric underneath the mark is contested, the gap between the two becomes the whole story. This is already visible in the most-watched private names: secondary buyers have at times paid valuations far above what disciplined primary investors will underwrite for the same company — a divergence we walk through in the 2026 private-market bifurcation, where a handful of AI leaders trade at 25–50x revenue while most of the market reprices downward. A wide gap between what the most aggressive buyer will pay and what a careful one will pay is precisely what a valuation debate looks like in practice.
The same scale of money is changing hands regardless: employees at the top AI companies have already cashed out roughly $14 billion of private equity through secondaries, well before any IPO. Abundant liquidity and a contested valuation can coexist — which is exactly why the question shifts from "can I sell?" to "what is the thing I'd be selling actually worth right now?"
A paper mark is not a realizable price
Two structural facts sit between a headline valuation and the cash you could realize, and both get larger when valuations are in dispute.
Share class. A company's valuation is usually set by its most recent preferred round. If you hold common stock sitting behind a large preferred stack, your per-share value can be a fraction of the headline — because preferred holders are paid first in a sale. This is the mechanism behind the last-round valuation myth: the $X billion belongs to the company, not automatically to your shares.
The secondary discount. Even for the right share class, a private buyer typically pays less than the last round to compensate for illiquidity, information gaps, and risk. That secondary discount widens when buyers are nervous about a valuation — and a public bubble debate is exactly the kind of thing that makes them nervous. How that clearing price is set relative to the last round is its own mechanic, covered in how private shares are priced.
For more on this topic
For a structured walk through the considerations behind the sell-or-hold decision, see our Shareholder IQ guide to whether to sell your startup shares.
What this means if you hold private stock
A valuation debate is uncomfortable precisely because it can't be resolved from the outside — no one can tell you whether your company's mark will hold. The useful response isn't to predict the outcome. It's to reduce what you don't know about your own position, so that whatever you decide is a deliberate choice rather than a reaction to a headline.
1. Know what you actually hold. Your share class and the preferred stack above it determine far more about your realizable value than the headline valuation does. This is knowable today, before any decision.
2. Know what it would clear at. The relevant number isn't the last-round valuation; it's what a real buyer would pay for your specific shares in the current market. When the headline is contested, the spread between those two figures is the information you're missing.
3. Weigh concentration on its own terms. If a large share of your net worth sits in one private stock, the valuation debate is really a concentration question. The issue isn't whether the company is good — it's whether that much of your wealth riding on one uncertain mark fits your situation.
4. Remember a partial sale exists. Because secondary liquidity now recurs at leading companies, selling a portion — taking some risk off the table while keeping upside — is a real option, not a hypothetical. It is one way to act under genuine uncertainty without betting everything on a single outcome.
5. Account for timing and tax. When and how you sell affects holding-period treatment and your broader tax picture, and depends on circumstances no article can know. This is not tax advice. Consult a qualified tax professional before you sell private company shares. For a general overview, see tax considerations in a secondary sale.
⚠️ A contested valuation is not a sell signal, and a rising one is not a hold signal. The debate doesn't tell you what to do with your shares — it removes the comfort of treating the headline number as settled. What to do with that is a question about your share class, concentration, tax position, and timeline, not about whether the market currently calls AI a bubble.
The takeaway
The bubble-or-not argument will run for a long time, and no one writing about it — including us — knows how it resolves. But you don't need to resolve it to act sensibly. The debate's real contribution, for a shareholder, is to make explicit something that was always true: the valuation marking your private shares is an estimate, and a particularly contested one when it rests on an inflated-ARR argument.
That reframes the work in front of you. The question stops being "will the AI valuations hold?" — which you can't control — and becomes "what do I actually own, what would it clear at today, and how concentrated am I in one uncertain number?" Those you can answer. They're the ground a deliberate decision stands on, whichever way the broader debate goes.
Have questions about what your private shares are worth?
We're happy to share our perspective.
If you're holding private stock and trying to separate the headline valuation from what your shares would actually clear at, we'd encourage you to reach out. We're always glad to talk through how this market prices private equity — no commitment required.
Get in touch →contact@earlyasset.com · We typically respond within one business day
Earlyasset, Inc. does not provide investment advice and is not a registered investment adviser. Pricing estimates are algorithmic and do not constitute an offer to buy or sell securities. All transactions respect the company's right of first refusal (ROFR) and any transfer restrictions in your equity agreements. Direct liquidity is provided by Earlyasset Capital, LLC, a separate entity from Earlyasset, Inc.
Source note: This article references a public StrictlyVC discussion published by TechCrunch on whether AI valuations and ARR are inflated, and related TechCrunch reporting on AI revenue metrics, based on public reporting as of June 2026. Company valuation and revenue-multiple figures referenced here come from the linked Earlyasset Research pieces and the public sources cited within them. Earlyasset is not affiliated with the investors, firms, or companies named in those sources and has no non-public information about them. Figures are as reported and may be revised. Shareholders considering any secondary sale should consult their own legal, tax, and financial advisors.
Tax disclaimer: This article is general educational content only and does not constitute tax, legal, or financial advice. Tax treatment of secondary transactions varies significantly based on equity type, holding period, state of residence, individual circumstances, and other factors. Consult a qualified CPA, tax attorney, or financial advisor before making any transaction decision. Earlyasset, Inc. is not a tax advisor and does not provide tax guidance.