September should have been a busy month for US IPOs. Instead, only four operating companies debuted (excluding SPACs), several pulled their offerings, and the two largest AI labs both reportedly pushed their timelines back. The Nasdaq hit record highs over the same period, so this isn't a market-wide sell-off. It's a public market being selective about price, disclosure and timing.

OpenEvidence started in 2021 essentially as a search engine built for doctors. It is trained on licensed medical literature and made available free to any licensed clinician in the US, with access verified by NPI number or a hospital email. Founder Daniel Nadler had already sold one AI company, Kensho, to S&P Global and he built this one on the same bet: a narrow model grounded in the right data beats a general one in a field where being wrong has heavy costs.

Clay was founded in 2017. It didn't cross real revenue until 2022. It hit $30 million in 2024, roughly $100 million in annualized revenue by last December, and an estimated $150 million by this May. Along the way its valuation went from $500 million to $1.25 billion to $3.1 billion to $5 billion in about 18 months. Seven years of wandering, then an 18-month sprint that, if you didn’t know any better, looks like an overnight success.

AI-assisted coding has changed that calculus. The time from idea to deployed application has compressed in ways that were genuinely difficult to anticipate even two years ago. And when software creation moves that fast, the plumbing that supports it starts to matter differently. Infrastructure that used to be a one-time decision is now being provisioned thousands of times a day, often by automated tools rather than human developers. That shift is where Supabase's story begins

SpaceX prices its IPO Thursday, with first trades Friday on the Nasdaq under ticker SPCX, at a fixed $135 per share. The base raise is $75 billion, rising toward $86 billion if underwriters exercise the over-allotment, the largest offering in history by a wide margin. In the same stretch, Anthropic and OpenAI each confidentially filed draft S-1 paperwork with the SEC. Anthropic filed first, on the heels of a funding round valuing the company near $965 billion. OpenAI followed

When we last wrote about Cerebras, the company had confidentially refiled for IPO and the debate was theoretical: would public markets reward a capital-intensive AI infrastructure company on strategic positioning alone, as longtime backer Coatue argued, or would “price matter” once a real-time pricing engine got involved?

For years, speech technology largely lived at the edge of the enterprise stack: call transcription, dictation, basic IVR, and narrow automation. That is changing. The latest data suggests enterprises are no longer treating voice as a feature. They are beginning to treat it as a core interface — and, increasingly, as a strategic layer of software.

Few private companies attract the level of sustained institutional attention that SpaceX does — and for good reason. Across the dimensions that matter most to sophisticated investors evaluating late-stage private opportunities — market position, financial profile, and path to liquidity — SpaceX presents a compelling case. It also carries at least one meaningful new variable worth watching carefully.

This week The Information reported that Cerebras Systems, the Silicon Valley chip designer built around its wafer-scale computing architecture, has confidentially refiled for a U.S. IPO that could come as soon as April 2026.

The IPO market is technically open…but it’s already becoming choosy. A week ago, Blackstone’s president declared that “2026 should be the year of the IPO.” Just days later, one of Blackstone’s own portfolio companies put that assumption to question. On the surface, it was a busy week for new listings, with seven IPOs and six SPACs pricing. But it was also notable for what didn’t happen.

Nvidia’s recent $20 billion acqui-hire of Groq is more than a single high-profile transaction — it’s a clear signal that consolidation in AI hardware is accelerating.