The Tech IPO Market in 2025-2026: CoreWeave, Klarna, and the AI Premium

The US tech IPO window was effectively closed from late 2021 through early 2024. The 2022-2023 rate environment, the SPAC hangover, and a deep markdown on private valuations meant most growth-stage companies stayed private and burned through secondaries. The 2024 trickle (Reddit, Astera Labs, Rubrik, Klaviyo on the late-2023 side) cracked it open. CoreWeave in March 2025 demonstrated the AI-adjacent premium. By mid-2026 the window is the most open it has been since 2021 — but it is selective.

This is the state of the tech IPO market through 2025 and into 2026 — who priced, who filed, who is still waiting, and what the AI-adjacent premium actually does to the math.

CoreWeave: the AI infrastructure IPO that reset the window#

CoreWeave priced its IPO in March 2025 at 40 dollars per share — the bottom of its 47 to 55 dollar range and below the prior week’s filing midpoint. The deal raised approximately 1.5 billion dollars at a roughly 23 billion dollar fully-diluted valuation. The pricing was disappointing relative to expectations; the aftermarket performance was not. The stock traded materially higher into Q2 2025 as the Nvidia GPU-cycle narrative reasserted, and CoreWeave’s revenue concentration with Microsoft (62 percent of 2024 revenue) became reframed as quality customer rather than concentration risk.

What CoreWeave demonstrated was that the AI-infrastructure narrative was bankable in public markets even at high leverage and concentrated customer exposure. The follow-on IPOs — Cerebras filed in late 2024, Astera Labs (which priced March 2024) continued performing, and the GPU-cloud and AI-chip cohort opened up — borrowed the CoreWeave framing.

The CoreWeave-Core Scientific deal in mid-2025 was the other story. CoreWeave’s acquisition of its largest hosting partner consolidated the GPU-cloud-to-data-center stack and removed a counterparty risk that the IPO prospectus had flagged. The deal’s structure (stock plus assumed debt) was the kind of move only a publicly traded company can do at that scale.

Klarna: the long-delayed European unicorn IPO#

Klarna’s IPO finally priced in late 2025 after years of speculation. The Swedish buy-now-pay-later company had been the European tech unicorn most associated with the 2021 frenzy — peak private valuation 45.6 billion dollars in mid-2021, a 6.7 billion dollar valuation in mid-2022’s down round, and a slow recovery through 2024 as the business returned to profitability. The IPO priced at a valuation in the high teens of billions, well below the 2021 peak but materially above the 2022 trough.

The post-IPO trading has been the more interesting story. Klarna’s positioning as an “AI-first” company — the Sebastian Siemiatkowski “AI replaced 700 agents” claim that became the most cited example of generative-AI labor substitution in 2024 — was walked back in 2025 as the company quietly rehired for customer service roles. The IPO marketing emphasized AI productivity gains; the post-IPO disclosures emphasized agent quality. The market priced the gap.

Stripe: filing speculation that did not become a filing#

Stripe has been the most-anticipated US tech IPO since 2021. The company conducted a tender at a 70 billion dollar valuation in early 2024 — down from the 95 billion peak but materially above the 50 billion 2023 mark. Subsequent secondaries through 2024 and 2025 priced at 91.5 billion and then north of 100 billion as the business compounded.

Through 2025 and into 2026 the IPO speculation has been continuous but the filing has not come. Stripe’s stated reasons have included regulatory uncertainty, founder preference, and the availability of secondary liquidity for employees and early investors. The realistic interpretation is that as long as Stripe can run tenders at rising valuations and fund operations from cash flow, the marginal benefit of going public is modest relative to the disclosure and governance overhead.

If Stripe does file in 2026 or 2027, it will be the largest tech IPO since Facebook. The market is pricing roughly even odds for a 2026 filing as of mid-year.

Discord: the consumer social IPO question#

Discord’s S-1 filing came in early 2026 — the consumer-social IPO the market has been waiting on since Microsoft’s 2021 acquisition attempt fell through. The business has 200 million monthly active users, strong gaming-and-creator engagement, and the long-running question of whether the freemium-plus-Nitro monetization model scales to public-market expectations.

The IPO marketing has emphasized the AI-and-agent angle — Discord as a platform for AI bots and creator-driven AI experiences, with the existing developer ecosystem as the moat. Whether the public market buys the AI framing or prices Discord as the consumer-social comp set (Reddit’s post-IPO performance is the most cited reference) will set the tape for the 2026 consumer-social IPO cohort.

Databricks: the 2026 IPO everyone expects#

Databricks has been the most consensus-expected IPO for 2026. The company raised at 62 billion in late 2024 — a Series J that closed alongside the Tabular acquisition — and has been telegraphing public-market readiness through 2025. The actual filing timing has slipped multiple times.

The Databricks IPO will be the most-watched data-platform listing since Snowflake. The comparable matters: Snowflake priced September 2020 at 120 dollars per share against a target of 75 to 85, doubled on day one, and reached a peak fully-diluted valuation north of 100 billion before the 2022 multiple compression. Snowflake’s 2025-2026 trajectory — under pressure on warehouse pricing, regaining momentum on the AI-and-Cortex angle — is the live comp for Databricks’s public-market positioning.

Our own Snowflake versus Databricks versus BigQuery post for 2026 covers the platform-side dynamics. The IPO will set a public-market data point on what enterprise data platforms are worth in the AI era.

The SPAC tail: Joby and the survivors#

The 2020-2021 SPAC wave produced hundreds of listings, most of which traded poorly through 2022 and were either acquired, delisted, or recapitalized. The survivors are an interesting cohort. Joby Aviation, the electric vertical-takeoff company, is the most visible: still publicly traded, deepening commercial agreements with Delta and Uber, on a path to commercial launch in 2026. The SPAC vehicle that took Joby public in 2021 looked like a 2021-era anachronism by 2023; by 2026 the company has more operational maturity than most direct-IPO listings of the same vintage.

The lesson is not that SPACs were a good structure — they mostly weren’t — but that survival depends on the underlying business, not the listing mechanism.

Reddit and Cava: the 2024 success stories that mattered#

Reddit’s March 2024 IPO at 34 dollars per share priced well and traded better. The stock more than doubled through 2024 and 2025 on a combination of subscriber-driven advertising growth, the OpenAI and Google data-licensing deals, and the broader rotation into ad-supported AI training data. Reddit became the second-most-cited example (after CoreWeave) of the AI-adjacent IPO premium.

Cava, the Mediterranean fast-casual chain, was the non-tech success story that also opened the window. June 2023 IPO at 22 dollars, immediate doubling, and sustained outperformance demonstrated that the IPO window was opening for businesses with strong unit economics regardless of sector. Cava’s success made underwriters more willing to bring 2024-2025 deals.

The AI-adjacent IPO premium#

The most-discussed phenomenon of 2025-2026 has been the AI-adjacent IPO premium — the consistent observation that companies with credible AI narratives priced and traded materially above sector comparables. CoreWeave at 4 to 5x next-twelve-months revenue. Astera Labs at multiples that would have been laughed at in 2023. Cerebras IPO marketing that explicitly priced against the AI infrastructure cohort rather than the semiconductor cohort.

The premium has limits. It rewards credible AI exposure (CoreWeave, Astera, the data-platform names) and punishes AI-washing (several mid-2025 IPOs that emphasized AI in marketing without underlying revenue substance traded poorly). The mid-2026 market is pricing the discrimination — AI-real gets a multiple, AI-claimed gets the sector comp.

For founders considering a 2026 or 2027 IPO, the practical implication is that the AI narrative must be defensible in revenue and product terms, not just in marketing. The premium is for AI businesses, not for businesses with AI slides.

Where pdpspectra fits#

We help enterprises and pre-IPO companies build the AI and data infrastructure that supports the growth narrative — production AI systems, data platforms, and the engineering scaffolding that underwriters actually examine in S-1 diligence. Our business automation practice does this work.

Related reading: Snowflake vs Databricks vs BigQuery 2026, datacenter power constraints in 2026, and tech M&A deals 2025-2026.

---

The IPO window is open and selective. Talk to our team about the AI and data infrastructure work that supports a credible public-market narrative.