The question sounds simple: how does a company go public? The answer involves a regulatory filing, a roadshow, a pricing meeting, a first day of trading, and a lockup period, and this week gave us three live examples to learn from. Nvidia-backed AI cloud provider Nscale filed to go public while losing more than a billion dollars in six months. Intel-backed chipmaker Altera filed confidentially for an IPO reportedly seeking to raise $2 billion. And SpaceX shares rose on Thursday even as a large slug of shares was freed for trading.
Let us use all three to explain the machinery.
First, what an IPO is
An initial public offering is the first time a company sells shares to the general public. Before the IPO, the company is private: its owners are founders, employees, and investors like venture capital firms. After the IPO, anyone with a brokerage account can buy in. The company raises money by selling new shares, and the early owners get a public market price for what they hold.
Companies go public through a process regulated by the Securities and Exchange Commission. The key document is the S-1, a registration statement where the company lays out its business, its financials, its risks, and how it plans to use the money. Think of the S-1 as the company’s full confession, written in legal language, filed for the world to read.
Story one: Nscale, the billion-dollar loss headed for the market
Nscale, an AI cloud provider backed by Nvidia and other investors, filed to go public this week reporting a $1.02 billion net loss on $140.6 million in revenue for the six months through June, according to Morning Brew’s September 19 edition.
Read that again. A billion dollars lost in half a year, on $140.6 million of revenue. And the company is asking public investors to buy in.
Why would a money-losing company go public? The honest answer has a few parts. First, growth companies often lose money on purpose, spending heavily to build data centers, hire engineers, and win customers before competitors do. The bet is that scale turns losses into profits later. Second, an IPO raises fresh capital to fund that spending. Third, it gives early investors and employees a way to eventually sell their shares.
The Nscale filing is also a window into the AI infrastructure frenzy. The company is building the cloud capacity that AI needs, and investors are currently willing to fund enormous losses to own a piece of that buildout. Whether that willingness lasts is the three-trillion-dollar question hanging over the whole AI trade.
Story two: Altera, the confidential filing
Intel-backed chipmaker Altera confidentially filed for a US IPO this week, reportedly seeking to raise $2 billion, according to The Daily Upside’s September 16 edition. The backstory is a lesson in private-market math. Intel bought Altera for $16.7 billion in 2015. Silver Lake later bought a 51% stake at roughly half that valuation. Altera’s 2024 revenue was $1.5 billion, down 48%.
A “confidential filing” means the company submitted its S-1 to the SEC privately, without the public seeing it yet. The SEC allows this so a company can work through regulators’ questions and test investor appetite without its financials becoming front-page news. If the process goes well, the filing becomes public later and the IPO proceeds. If not, the company can quietly walk away.
Altera’s numbers tell a turnaround story in progress: revenue nearly halved to $1.5 billion in 2024, the valuation cut roughly in half from Intel’s 2015 purchase price, and now a public offering aimed at raising $2 billion. Public investors will get to decide whether the worst is behind it.
The middle of the process: underwriters, the roadshow, pricing
Between the filing and the first trade, three things happen.
Underwriters, usually investment banks, guide the company through the process and commit to selling the shares. They help set the price range, line up buyers, and stabilize trading in the early days.
Then comes the roadshow. Company executives travel, in person and virtually, to pitch institutional investors: pension funds, mutual funds, the big buyers. This is where demand gets measured. Strong demand pushes the expected price up. Weak demand pushes it down.
Then pricing. The night before trading begins, the company and its underwriters set the final offering price. This is the price the chosen investors pay. Here is the part retail investors often miss: by the time you can buy on the first morning, the offering price is history. You are buying at the market price, which is whatever traders decide that morning.
Story three: SpaceX and the unlock
SpaceX shares rose Thursday even as a large slug of new shares was freed for trading, according to Axios. The company’s June IPO was structured so that less than 5% of shares could initially trade, which constrained supply.
This is about lockup periods. After most IPOs, insiders and early investors agree not to sell for a set time, often 180 days. The lockup prevents a flood of selling on day one. When the lockup expires, the “unlock,” a wave of shares becomes eligible for sale, and the stock often falls on the extra supply.
SpaceX did it differently. By limiting the initial float to less than 5% of shares, the company kept supply tight from the start. So when more shares were freed for trading on Thursday, the stock still rose. Scarcity did its work.
The lesson is general even though the example is unusual. Supply and demand do not pause for IPOs. A great company with too many sellers at once can see its price sag, and the unlock calendar is one of the first things a careful IPO investor checks.
The first day, and what comes after
On the first day of trading, the stock opens at whatever price the market sets, often above or below the offering price. A “pop,” where the stock jumps on day one, makes headlines and makes the offering look hot. But here is the part worth saying plainly: a first-day pop is not a measure of the company’s long-term value. It is a measure of one morning’s enthusiasm. Plenty of stocks that popped on day one have drifted for years. Plenty that stumbled on day one became great investments.
A note for the regular investor
If you are a regular person with a brokerage account, here is the honest truth about IPOs. Getting shares at the offering price is hard. Those shares go to the underwriters’ best institutional clients first. By the time you can click “buy,” you are paying the market price, and the easy first-day gains, if any, may already be gone.
That does not mean you should never buy a newly public company. It means you should buy it the way you would buy any stock: because you have read the filing, you understand the business, and you believe in the price you are actually paying. The S-1 is public. The numbers are there. Nscale’s billion-dollar loss, Altera’s halved revenue, SpaceX’s tight float: all of it was knowable before anyone traded a share.
Three companies, three different doors into the public market. One losing money on the way in, one quietly testing the waters, one rewriting the rules of supply. The machinery is the same underneath. Now you know how it works.
