Silicon Valley has long leaned on the “Onion Theory of Risk” — coined by Andy Rachleff and popularized by Marc Andreessen. It frames a startup as layers of risk: team, product, market, scale. As each layer is peeled away, the company de-risks and value is created. It’s an optimistic story — and it’s told entirely from the founder’s and the early investor’s side of the table.
Olympus studies the other side of that onion. Because every layer of risk peeled away isn’t just value created — it’s value claimed. And the people who claim it first are insiders. By the time a company reaches the one door most investors can walk through — the public market — the richest layers are already gone. What’s left is the core. The Onion Theory of Exit Risk maps who took which layer — and what the core is really worth by the time it lands in your hands.
The Five LayersThe value-extraction sequence
The Founder Layer
Founder secondaries — the pre-IPO slice. Founders sell tranches in late private rounds and company tender offers, banking generational wealth years before you’ve heard the ticker. (The bigger founder cash-out comes after the IPO, when lockups expire — that’s Layer 5. The S-1 reveals the whole setup: who owns what, who’s selling into the offering, and exactly when the lockups lift.)
The Early-Capital Layer
Seed and Series A funds win two ways — both private. Every later round at a higher price marks up their stake, so they report gains to their own investors without selling a share. Then they sell part of the position through secondaries and SPVs at 10–100×, banking realized returns long before any prospectus is filed.
The Crossover Layer
Late-stage and crossover funds bid the valuation up round over round, and mutual funds mark the position — manufacturing the growth narrative you’ll later read in the press. Sometimes the peeling ends right here: insiders sell to each other in a private deal, and the public is never invited (see Ampere, below).
The Pre-IPO Retail Layer
SPVs, forward contracts, and “pre-IPO access” platforms finally let retail in — at a premium, with management fees and carry buried in the structure. The first time you can buy, you’re already buying someone else’s markup.
The Public Core
The IPO. Retail buys freely at last — at peak valuation, into a thin float and a wall of insider shares waiting behind lockups. When those lockups expire, supply floods and the core deflates. Most people call this layer “getting in early.” It’s usually the last layer, not the first.
Reading the CoreThe two forces that decide what happens next
Here’s the part most cautionary tales miss: the Onion Theory doesn’t say “retail always loses.” It says read the core honestly. Once a company is public, two opposing sets of forces decide whether that core deflates or holds. Knowing which is winning — and when — is the whole game.
Lockup expirations — the supply flood
The single most mechanical risk. When insider shares unlock, sellers who bought layers ago meet buyers at today’s price. Watch the calendar, not the hype.
Free float & the index distortion
A giant headline market cap with only a small slice publicly tradable creates a dilemma for index providers: weight it at full economic value and you force hundreds of billions in distorted reallocation that can pump the price artificially — or weight it by actual float, and index investors get far less real exposure than the headline suggests. Either way, thin float plus locked supply means a core that can swing violently when shares release.
Profitability → S&P 500 inclusion
To join the S&P 500, a company generally needs four consecutive profitable quarters (positive trailing-four-quarter and most-recent-quarter earnings), among other criteria. Clear that bar and index funds become forced buyers — a powerful, mechanical demand catalyst that can re-inflate the core.
Corporate profits after tax — the macro tide
Aggregate after-tax corporate profits are the tide under every valuation. Expanding profits widen risk appetite and support multiples; contracting profits pull liquidity out and let cores deflate faster. Every exit decision sits on top of this gauge.
When the Revenue Is the Customer
Cerebras’ IPO was textbook euphoria — it priced above range, valued the company near $23B (roughly 90× sales), and the stock more than doubled. But read the core. Growth was overwhelmingly concentrated in two UAE-affiliated customers (MBZUAI and G42, together ~86% of 2025 revenue), and its $24.6B backlog leans heavily on a single name: OpenAI. G42 didn’t just buy compute — it invested $335M, a customer-investor tangle tight enough to draw a CFIUS review and force the first IPO filing to be withdrawn.
Then the first public earnings landed (June 23, 2026) — and the framework played out in real time. Revenue beat: ~$191M, up 92%. But management guided next quarter’s gross margin down to 36–38% and operating margin to roughly negative 30% — squeezed by the pass-through data-center costs of the very OpenAI buildout that headlines the backlog. The prospectus had warned those costs would be “dilutive” to margin. On the print, the stock sold off.
Strong headline, pressured core. The framework tells you to ask not just “how fast is revenue growing?” but “whose money is it, at what margin — and would it still be here if they weren’t also an owner?”
A Core With No Floor — Yet
Quantinuum — the trapped-ion leader spun out of Honeywell and Cambridge Quantum — priced its June 2026 IPO at $60, above range, raising $1.68B at a ~$15B valuation. Note the inversion: this was not an insider dump. The offering was primary (the company raised the cash), and Honeywell kept ~49% voting control. Nobody peeled the onion on the way out.
So where’s the exit risk? In the price itself. Quantinuum earned ~$31M in 2025 against a ~$193M net loss — an IPO valued at 450×+ revenue, betting the company reaches fault-tolerant quantum (a ~2029 timeline) before anyone else. With no profit floor to catch a fall, the stock did exactly what an un-anchored core does: it dipped below its $60 IPO price, then ripped back to the high-$60s within weeks. That’s not value discovery — it’s sentiment swinging on a core the market can’t yet price.
The lesson: sometimes the exit risk isn’t who’s selling — it’s that the core has no fundamental floor, so it lives and dies on the story.
The Live OnionWorked examples
Frameworks are only useful when you can point them at something real. Two cases — one heading for the public market, one that never got there.
Peeling the SpaceX Onion
The early layers are long gone. Founder, early-VC, and crossover value was claimed across years of private rounds and secondaries on the way to a $1.75T valuation. Retail’s first real access came through pre-IPO SPVs — at a premium. Now the public core arrives, and the framework flags both forces at once, on a known calendar:
A low free float against an enormous headline cap — and a lockup schedule that, by Olympus’s read, leaves only ~5% of shares unlocked through October, with ~60% freeing up in late fall. That release is the sternest test of the stock’s resilience — a defined window where exit risk spikes.
Profitability trajectory. With Starlink monetizing aggressively — the Google partnership, multi-billion-dollar Amazon agreements — and a launch moat no rival is close to matching, consecutive profitable quarters put S&P 500 inclusion in play. String four together and forced index buying could overwhelm the lockup supply.
That’s the point of the framework: SpaceX isn’t simply “buy” or “avoid.” It’s a core being pulled by two mechanical forces with a known calendar. Read it right and you can position around the lockup window and the inclusion catalyst. Read only the headline and you’re buying the depleted core at the top.
Get the full SpaceX deep-dive brief — $19 →The Onion That Never Reached You
Ampere Computing built ARM-based data-center chips with Oracle as both a major backer — a ~32% stake, roughly $1.5B invested over the years — and a flagship customer. A 2023 IPO was floated, then shelved. So how did the insiders get their money out?
In 2025, SoftBank acquired Ampere outright for $6.5 billion in cash. Oracle and Carlyle sold their positions in the deal — Oracle put in ~$1.5B and walked away with nearly $1.9B. Every layer of that onion was peeled in private, between insiders. Retail never got a single share.
That’s the exit risk almost no one warns you about: the most valuable cores are often sold strategic-to-strategic before a ticker ever exists. If you’re waiting for the IPO, you may be waiting for an exit that never comes — or that arrives only after the real money has already changed hands.
See the Whole OnionBefore you buy the core
You can’t stop insiders from peeling the onion first. But you can refuse to buy blind. Every Olympus Intelligence Brief maps the layers that actually decide your outcome:
- The cap table and secondary activity — who already took their slice
- Real revenue vs. the manufactured narrative
- Free float, lockup calendars, and index-inclusion math
- A plain-English conviction rating on whether the core still has value
Before the IPO. Before Wall Street prices it in.