SpaceX is expected to come to market this year. In February 2026, SpaceX acquired xAI outright in a deal that valued the combined entity at $1.25 trillion, and the entire xAI risk surface now sits inside the SpaceX corporate perimeter. Anthropic and OpenAI are also expected to file. The S-1s, when they land, will be among the most heavily read prospectuses in a generation. Most of the attention will go where it usually goes: revenue concentration, GPU supply, model-quality disclosures, the headline regulatory risks. The sentences worth reading carefully will be elsewhere. They will be the ones about pending litigation, training-data provenance, and intellectual property indemnities. Somewhere in those filings will be a sentence that, twenty years from now, the retrospectives will quietly leave out.

The investor lens

I know what to look for because I have seen this pattern before, once from the inside, in a way that almost no Australian investor lived through directly.

Founders sometimes ask me what an angel investor brings to the table beyond the cheque. The honest answer varies by angel, and that variation matters more than people sometimes acknowledge. Some bring sector depth. Some bring distribution. Some bring deep operational experience in a particular function. As a former CEO, I can engage on operational issues when they come up. But the contribution I’m best positioned to make, and the one I think mattered most at Dark Blue Sea twenty years ago, is what I think of as the investor lens.

The lens isn’t glamorous. It’s pattern recognition built up over nearly forty years in and around capital markets: the trained instinct for how disclosure regimes, prospectus liability, listing rules, and surprise legal events can warp the financing path of an otherwise healthy company. Most early-stage founders don’t have it, and as a general rule I don’t expect them to. The two skills I do look for are someone who can build the product and someone who can sell it. Finance capability is a bonus, not a requirement. The lens is what I’m there to provide.

Patents are one of the categories where the lens earns its keep. They are a particularly fertile source of surprises. Granted is not the same as valid. Strong patents can turn out to be invalid; weak patents can be commercially decisive. I have argued in another post that, for a startup, a strong patent is the only form of IP that creates real structural advantage. The corollary is the harder lesson: someone else’s patent can close doors you didn’t know were open.

The GoTo patent: a case study in disclosure risk

The case study is the early-2000s pay-per-click patent saga. Most readers will know about it through Google. Hardly any will remember what it did to Google’s IPO, and almost none will know what it did to Dark Blue Sea. A patent litigated in Pasadena ended up closing the IPO door on a company in Brisbane without a single piece of correspondence ever crossing the Pacific. I was the CEO of the Brisbane company.

In February 1998, Jeffrey Brewer of GoTo.com presented a pay-per-click search engine proof-of-concept at the TED conference in California. By July 1998, advertisers were paying anything up to a dollar per click. The service was public, commercial, and well known throughout 1998 and into 1999. (I’ve written elsewhere about how, two years later as CEO of Dark Blue Sea, I came to see this business model up close, and why it convinced me the commercial internet was real even when most people had given up on it.)

GoTo did not file a patent during any of that period. It got around to the patent process in May 1999, by which point, on Gross’s own subsequent admission, it had already lost the ability to patent the core invention. US patent law gives an inventor one year from first public disclosure to file. February 1998 to May 1999 is fifteen months. The horse had bolted.

What prompted the May 1999 filing was the IPO. Gross’s own account, given to Steven Levy and reproduced in In The Plex, is unusually candid:

“We were ready to go public … and were getting our IP portfolio together for the bankers, and everybody was like, ‘What patents do we have?’ And we didn’t have too many. … All we could do was patent everything else we could think of, a bunch of obscure things like the way we accept bids. These were silly patents, but the real patents would have been worth billions.”

This is a remarkable thing for a founder to have said in print. The patent that would, three years later, hold up Google’s IPO and reshape the financing options of every small PPC operator in the world is described, by the man who filed it, as a silly patent assembled because the real ones were no longer available.

The patent, US 6,269,361, was nonetheless granted in July 2001. By then GoTo had renamed itself Overture Services and was syndicating PPC inventory to Yahoo, MSN, AOL and most of the major portals. And once the patent issued, Overture moved.

The first major target was FindWhat.com, a Florida-based PPC operator, with litigation beginning in January 2002. Overture then sued Google in April 2002. The validity case turned on exactly the issue Gross had publicly acknowledged: that the technology had been in commercial use more than a year before the patent was filed. The campaign broadened from there. Lycos was added in 2004. Cease-and-desist letters went to other operators. The message to the small-cap PPC world was clear: the largest player held a granted patent and was prepared to litigate. Whether the patent would survive a full validity challenge was a separate question; defending an infringement action would still cost millions of dollars and several years.

When Google filed its S-1 with the SEC on 29 April 2004, the Overture suit was disclosed as a risk factor. The phrase that mattered, given Google’s revenue base, was that an adverse outcome could “significantly limit our ability to use the AdWords program.” Google’s own filings showed that AdWords advertising accounted for 99% of revenue. The S-1 disclosure was, in substance, an admission that 99% of revenue was contingent on a litigation outcome, and the parties’ own pleadings showed that Google’s lawyers thought the patent was probably invalid for prior-use reasons. None of that produced an actual courtroom ruling. It produced a settlement.

The IPO was already in trouble for other reasons: a Playboy interview during the SEC quiet period, allegations of unregistered employee shares, a MyDoom virus variant hitting google.com, and broad institutional resistance to the Dutch auction format. But the unresolved Overture exposure was a structural problem in a way the others were not. It touched the revenue line, and it was unresolved at the level of valid versus invalid. Initial price guidance had been $108 to $135 per share. The range was cut to $85 to $95, and the IPO ultimately priced at $85, for a valuation of $23 billion.

On 9 August 2004, days before final pricing, Google and Yahoo announced they had settled the Overture lawsuit. Google issued 2.7 million shares of Class A common stock to Yahoo in connection with the settlement arrangement. Google received a perpetual licence to the ‘361 patent and several related patents, and recorded a non-cash charge of $260 to $290 million in the third quarter, producing a quarterly net loss. The IPO priced at $85 on 18 August. It closed the next day at $100.34.

The story since told about Google’s IPO is the Dutch auction, the Founders’ Letter, “don’t be evil,” and the first-day pop. The story not told, or told only as a footnote, is that an obscure patent assembled in fifteen months by a competitor’s IPO scramble forced a roughly $9 billion reduction in implied valuation, a 2.7 million share payment to Yahoo, and an admission in an SEC filing that 99% of revenue was contingent on a lawsuit. Twenty years of Google hagiography has quietly airbrushed that part away.

There is a footnote that deserves more attention than it usually receives. Yahoo’s settlement equity, combined with prior holdings from a 2000 services agreement, totalled roughly 8.4 million Google shares. Yahoo sold them into a rising market across 2004 and 2005 for a combined total of approximately $1.4 billion. The patent had been late-filed, arguably invalid, and described by its own founder as silly. Its enforcement nonetheless converted itself, via the settlement, into a backdoor equity play on Google’s success, the very competitor that was, in the same window, destroying Yahoo’s search business. Yahoo’s IP victory was complete. Its operational defeat was concurrent and absolute.

The Dark Blue Sea story nobody knows

The Dark Blue Sea version of the story is less visible but, from where I sat, more instructive. DBS was an unlisted Australian internet company. It owned a pay-per-click search engine. Bid-for-placement PPC accounted for more than 90% of revenue. The entire business sat squarely within what the ‘361 patent purported to cover.

We had been actively considering an IPO. The growth was there, the metrics worked, and an ASX listing was the natural next step.

The Overture enforcement campaign closed that door from the inside.

No letter ever arrived. None had to. The disclosure regime did the work. Any conventional IPO would have required substantively the same Overture-patent risk disclosure that Google had been forced to make. For DBS, that disclosure would have read substantially worse than Google’s: 90%-plus of revenues subject to a third-party patent claim, in a company without Google’s resources to defend an action and without Google’s settlement currency. No Australian underwriter or institutional investor would have engaged with that risk profile in 2002 or 2003. The IPO died, not because we were sued, but because the disclosure obligation rendered the financing path uneconomic before any letter could arrive.

We did two things in response. First, we reached the ASX via a reverse listing instead of an IPO, a structurally different transaction with a much lighter prospectus-style disclosure burden, which still delivered public-market liquidity, a share register, and a platform for subsequent capital raisings and acquisitions. Second, we redesigned the business around the edges of the patent claims, engineering a model that operated outside the bid-for-placement mechanics the patent purported to cover.

Looking back, we did better with this than Google did. Not because we were smarter on the law; we weren’t, and Google had access to the best patent lawyers in the world. Different problem, better tools. Google’s lawyers were fighting an active suit on the merits. Our investors were redesigning the financing path before a letter ever arrived. Someone in the room had seen patent disputes warp financing paths before, recognised the disclosure problem before it became a litigation problem, and routed around it. That is what the lens does.

Back to the launch pad

Which brings us back to the S-1s on the launch pad.

AI feels a lot like the early days of the internet boom. Same sense of acceleration. Same gap between commercial momentum and legal infrastructure. Same race to file patents, some real, many silly. The disputes have already started: at the model layer, the data layer, the application layer. Plus copyright, training-data provenance, and a dozen legal categories that didn’t exist a few years ago. (I’ve argued elsewhere that the AI cycle parallels the dot-com era in important ways. The patent cycle is one more parallel I expect to play out.)

The Anthropic and OpenAI S-1s, when they file, will need to characterise this thicket. Both companies have active and material litigation exposure on training-data and intellectual property questions. Read the risk factors slowly. The phrasing matters. Somewhere in those filings will be the modern equivalent of “significantly limit our ability to use the AdWords program.” It will be a sentence the company’s lawyers would rather not have written and the bankers would rather you skim past.

The SpaceX filing will be the one to watch most closely, and it is the configuration that has no real precedent.

SpaceX acquired xAI outright in February 2026, making it a wholly-owned subsidiary. That means its risk surface, including any litigation involving xAI’s competitors that touches xAI’s strategic position, flows directly into the SpaceX S-1. Musk is seeking to have Altman and Brockman removed from their roles and to unwind the restructuring that allowed OpenAI to operate a for-profit subsidiary, in a trial whose outcome could upend OpenAI’s race toward an IPO. The SpaceX prospectus will have to disclose all of this: the suit, the relationship of the litigant to the issuer, the strategic position of the subsidiary whose competitor is being sued, and the contingent outcomes. There is no neat way to write that paragraph.

The 2004 parallel is striking, but the 2026 configuration is sharper. Overture sued Google because it wanted Google’s money. Musk is suing OpenAI for reasons that the litigation papers describe in existential terms, and the IPO that must disclose the suit is the IPO of the company that owns Musk’s competing AI business. The disclosure obligation and the strategic objective sit on the same page in a way that, to my knowledge, no prior major IPO has ever had to navigate.

I have no idea how the lawyers will draft that risk factor. I do know it will be the most carefully read paragraph in the prospectus, and that the language they choose will be in retrospectives of this market for a long time. Read it slowly.

In a decade, similar stories will be told. Some founder will discover that an obscure patent filed in fifteen months by someone else’s IPO scramble has reshaped their financing options. Some IPO will be repriced. Some company will quietly redesign around the edges of someone else’s claims. Some retrospective will airbrush the legal mess out of the founding narrative. The specifics will change. The pattern will not.

So when a founder asks what I bring to the table beyond a cheque, the honest answer is the lens. Build the product. Sell the product. Bring an investor whose pattern recognition catches the disclosure problem before it becomes a litigation problem, and who has seen, from the inside, what capital markets can do to a good idea. Other angels bring other things. This is what I bring.

It turns out to have been the most useful thing I did at Dark Blue Sea. I expect it will be among the most useful things I do at MooCoo too, because the AI vintage is going to throw up its own version of the GoTo patent, and someone is going to need to see it ten miles before it arrives.

Richard Moore is co-founder of MooCoo Ventures, an angel syndicate that co-invests alongside Brisbane Angels, one of Australia’s most active angel groups. He has made over eighty personal angel investments since 2013.

moocoo.vc