The stock market loves a massive public offering, but few corporate prospectuses have landed with the sheer gravitational pull of Anthropic's S-1 filing. When a private lab turns into a public beast aiming for a valuation north of a trillion dollars, Wall Street scrambles to read the fine print. Everyone sees the staggering revenue growth. Few are looking closely at the massive compute obligations and governance structures hiding in the back pages.
If you are trying to figure out whether to buy in or run away, you need to strip away the hype. The numbers tell a wild story about modern infrastructure, capital burn, and who really calls the shots when the bell rings.
The Revenue Explosion Versus Reality
Let's start with the top line because it is genuinely jaw-dropping. Anthropic reported $16.3 billion in revenue for just the first half of 2026. Compare that to $4.6 billion across the entire previous year, and you are looking at growth that defies traditional corporate gravity. Claude is everywhere, businesses are plugged in, and enterprise contracts are scaling faster than anyone anticipated.
Gross margins have also ballooned to roughly 85%, looking more like a software monopoly than a heavy infrastructure player. But gross margins don't tell you the whole story. Strip away the surface sparkle and you find a first-half net loss of $80.7 billion.
Before you panic, understand what caused that red ink. The vast majority of that loss stems from non-cash accounting adjustments tied to convertible notes and valuation changes. It is an accounting quirk, not a pile of cash vanishing into thin air. Even so, running an artificial intelligence titan requires a terrifying amount of capital, and that brings us to the hardware bottleneck.
The Staggering Cost of Compute
You cannot talk about large language models without talking about silicon. Training and running frontier intelligence eats electricity and chips at a scale humanity has never witnessed.
Anthropic's filing lays bare a massive commitment of up to $120.5 billion directed toward compute infrastructure through 2029. This includes massive multi-billion dollar monthly outlays once full capacity kicks in. When people ask if the artificial intelligence boom is a bubble, this is what they are pointing at. The capital expenditure required to stay competitive is astronomical.
If revenue growth slows even slightly while compute costs continue their relentless climb, the cash crunch could arrive faster than expected. Investors are betting that software monetization will outpace hardware inflation. If compute expenses outstrip efficiency gains, those brilliant gross margins will compress in a hurry.
Who Is Really Running the Show
Public market investors like to think buying a share gives them a voice in the boardroom. Anthropic’s S-1 shatters that illusion completely.
The filing reveals a dual-control structure that leaves public shareholders with virtually zero directional influence. The seven co-founders hold a single Class F share through a dedicated entity that locks down 50.1% of the total voting power. Meanwhile, a separate trust elects the majority of the board.
You get the economic exposure, but the founders keep absolute control. If you disagree with their long-term vision, safety priorities, or commercial pivots, your only real option is to sell your stock. That kind of corporate autocracy is becoming common in tech, but it is worth keeping in mind before you allocate hard-earned capital.
What You Should Do Next
Do not buy the ticker symbol on Day One just because the revenue growth looks sexy. Read the risk factors section of the S-1, particularly regarding compute dependencies and user safety policy enforcement lag. Watch how enterprise retention holds up over the next two quarters. If you decide to take a position, size it like the speculative venture it is. Keep your eyes on the cash flow statements, ignore the noise, and remember that hypergrowth always comes with a heavy dose of volatility.
