Arm's IPO: scarcity is the selling point

SoftBank bought Arm for $32 billion in 2016 and now wants a valuation approaching $64 billion while selling just nine per cent of the British chip designer to the public. Masayoshi Son has always enjoyed a financial spectacle, and this listing is his most carefully engineered yet. Investors are hungry for semiconductor exposure, and scarcity is standing in for rapid sales growth.

From Cambridge to Nasdaq

The road here has been winding. SoftBank agreed the original purchase weeks after Britain voted to leave the European Union, when a weak pound made a British champion look cheap. In 2020 it agreed to sell Arm to Nvidia, and that deal collapsed in early 2022 after regulators in America, Britain and Europe objected. A flotation was the fallback, and Arm's choice of New York over London this spring made plain that the aim was the richest available price.

The reference price has also been set in-house. Last month SoftBank bought out the stake held by its own Vision Fund at a figure implying a value of about $64 billion for the whole company. That was a transfer between two pockets of the same group, with no outside buyer testing the number. It hands the underwriters an anchor that no independent investor has had to endorse.

Arm made its public filing with American regulators on 21 August, and the timing matters as much as the price. Listings in New York almost dried up in 2022 as rising interest rates hit technology valuations, and bankers have spent this year waiting for a deal large enough to reopen the market. Instacart and Klaviyo filed their own prospectuses days after Arm, betting that a successful chip listing will draw investors back. Arm is therefore being sold as a test of the whole new-issue market, which gives every underwriter and prospective issuer an interest in a strong debut.

A small float, a big price

By keeping more than ninety per cent of the shares, SoftBank limits the free float and creates a supply squeeze that forces index-tracking institutions to bid hard. Arm's fundamentals are less exciting. Its smartphone royalty base is mature, and its push into cloud data centres runs into established competition from proprietary chip designs. Investors are paying a premium multiple for shares that are strictly rationed.

The mechanics favour the seller. Fund managers measured against semiconductor benchmarks struggle to ignore a company whose designs sit inside nearly every smartphone, so part of the demand is close to compulsory. With so few shares on offer, a modest volume of buying moves the price a long way, and that price is then applied to the entire company, including the portion that never trades. The headline valuation is set by the keenest bidders for a small allocation.

Book-building reinforces the effect. Underwriters gather orders from institutions, and when a deal is heavily oversubscribed they allocate shares in small portions, often scaling back each order sharply. Investors who expect to be cut back inflate their orders, which makes demand look even stronger and justifies a higher price. Those who receive too few shares then buy more once trading starts, pushing the price up on the first day. All of this can happen without a single investor having to judge whether $64 billion is a sensible value for a business whose revenue is barely growing.

Royalties that grow slowly

Arm earns licence fees when chipmakers adopt its designs and a small royalty on every chip shipped. That model is steady but tied to unit volumes, and global smartphone shipments have been falling. Revenue in the year to March was roughly flat, which sits awkwardly with a price tag that assumes years of brisk growth. Arm is trying to lift the royalty per chip through its newer v9 designs, which asks customers with other options to pay more.

Those options are growing. Amazon's in-house Graviton server chips use Arm designs, but the bulk of data-centre processors still run on the x86 architecture controlled by Intel and AMD. The open-source RISC-V standard offers a royalty-free route that Chinese firms in particular have taken up. Arm is also suing Qualcomm, one of its largest customers, over licences Qualcomm acquired when it bought the start-up Nuvia. And a large slice of revenue comes from China through a joint venture whose control SoftBank sold in 2018.

The artificial-intelligence premium

The bull case rests on artificial intelligence. Nvidia's market value passed $1 trillion in May on demand for AI chips, and any plausible link to that theme now earns a premium. Arm's designs will sit in many of the processors that run alongside AI accelerators, and its reach across phones, cars and servers gives it plenty of options. The argument has merit, but Arm collects a small royalty per chip rather than the price of the chip, so the AI boom lifts its revenue far less than it lifts Nvidia's.

The best case for the price

The strongest argument for the valuation is that Arm's position is close to unique. Its instruction set is the standard for mobile computing, the software written for it represents years of investment by phone makers and app developers, and switching to an alternative is slow and costly. A business with that sort of lock-in can raise prices, and the move to v9 designs, which carry higher royalties, shows it intends to. If Arm's share of data-centre and car chips grows even modestly, royalties per device could rise for years while unit volumes stagnate.

That argument has force, but it describes a gradual story rather than a rapid one, and the price assumes rapid. Customers such as Qualcomm and Apple have the engineering muscle to resist higher fees, and the litigation with Qualcomm shows that Arm's biggest licensees will fight for their terms. The more Arm tries to extract from its dominance, the more attractive RISC-V becomes to customers looking for leverage in negotiations.

After the lock-up

Scarcity has a shelf life. Lock-up agreements usually stop insiders selling for six months after a listing, and SoftBank, after heavy losses at the Vision Fund, has every reason to sell more once they expire. When 3Com floated a small slice of Palm in 2000, the listed stub implied a value for Palm greater than that of its parent, and the anomaly closed once 3Com distributed the rest of its holding. The signs worth watching are Arm's first results as a listed company, whether higher royalty rates show up in revenue, and how much SoftBank chooses to sell when it can.

Arm's IPO shows off SoftBank's command of market mechanics more than any boom in chip sales. Once the float widens, the valuation will have to stand on its own, and a mature royalty base will struggle to carry it. The test is likely to come around the first lock-up expiry next spring, and the investors most exposed are the fund managers who bought because their benchmarks left them little choice, at whatever price the scarcity produced.