The IPO backlog is a myth
1. Key Themes
Theme 1: The "IPO backlog" is overstated, and the exit window may not release pent-up supply
Few of the "likely IPOs" actually listed
The thesis is that venture's expected wave of IPOs is smaller than assumed because most predicted exits never materialized.
"PitchBook's VC Time to Exit Predictor, which estimates when individual VC-backed companies are likely to exit, gave 254 US companies better-than-even odds of going public by the end of 2025." "Only 53 did, and just 58 still look likely to list in the next three years."
Companies are raising instead of listing
Backlog companies have found alternatives to going public.
"Nearly 75% of the backlog has raised new funding since 2023, mostly at higher valuations, and deals at Databricks and Scale AI paid out employees and early investors without a listing."
The scale of trapped capital
The size of the stakes explains why LPs are watching this.
"Venture has spent three years waiting for the IPO window to reopen, with $6 trillion now tied up in US unicorns alone."
Theme 2: AI is reshuffling which companies are IPO-relevant
Software incumbents face pressure while AI natives jump the queue
"AI has put pure-play software companies under pressure to prove their value, while a new generation of AI companies, led by Anthropic, has jumped to the front of the line."
Anthropic's S-1 validated outside estimates
The amended filing matched PitchBook's forecasts closely, lending credibility to their modeling approach.
"The revenue figures in Anthropic's amended S-1 match our estimates to the dollar. Q3 revenue came in at $16.96 billion against the $16.9 billion we implied from the original filing's disclosure, and commitments due before 2031 came in at $447.5 billion, just under the $450 billion line we set."
Theme 3: The AI data center buildout is channeling capital into unglamorous adjacent industries
The beneficiaries aren't software companies
"The AI boom's biggest beneficiaries may not write software at all."
Capital flowing into physical infrastructure and ancillaries
"A recent PitchBook research note maps how the mad dash to build more data centers is channeling private capital into areas such as gas pipelines, fire suppression systems, and even cargo freighters."
Investment has grown dramatically
"The combined VC, PE, and M&A investment across data center-exposed companies reached $64.4 billion in 2025, roughly 37x 2016's total."
Theme 4: Private markets keep absorbing capital through large raises and new credit and infrastructure vehicles
Large private-credit and infrastructure vehicles
"Apollo Global Management launched a €10 billion European private credit platform with backers including the Abu Dhabi Investment Council." "London-based Igneo Infrastructure Partners closed European Diversified Infrastructure Fund IV on its €6 billion hard cap." "Churchill Asset Management closed on $1.54 billion for the third fund in its Junior Capital Opportunities program, which invests in junior debt for middle-market businesses."
Early-stage venture funds still closing
"Stockholm-based VC firm Creandum closed its eighth fund, which backs founders from pre-seed to Series A, on $750 million."
2. Contrarian Perspectives
"The IPO backlog is a myth"
The consensus view holds that a backlog of mature startups will flood public markets once the window opens, returning capital to LPs. PitchBook argues the pipeline is much thinner than assumed.
"The hope is that when the window does reopen, a backlog of IPO-ready companies that have been patiently waiting for years will rush out in a wave and finally return capital to LPs. Our research argues that this 'IPO backlog' is little more than a myth."
The evidence is the gap between predictions and outcomes, plus alternative liquidity paths:
"Only 53 did, and just 58 still look likely to list in the next three years." "Nearly 75% of the backlog has raised new funding since 2023, mostly at higher valuations, and deals at Databricks and Scale AI paid out employees and early investors without a listing."
AI's biggest winners may not be software companies
Most investors frame AI as a software and model story. The newsletter argues the physical supply chain is capturing outsized value.
"The AI boom's biggest beneficiaries may not write software at all."
The supporting figure is the 37x growth in investment across data-center-exposed companies since 2016, reaching $64.4 billion in 2025.
3. Companies Identified
- Description: AI company that has filed an S-1 for an IPO.
- Why mentioned: Case study in both the IPO pipeline and the validation of PitchBook's financial estimates.
- Quotes: "The revenue figures in Anthropic's amended S-1 match our estimates to the dollar." and "a new generation of AI companies, led by Anthropic, has jumped to the front of the line."
- Description: Data and AI platform company.
- Why mentioned: Example of a company that provided liquidity without going public.
- Quotes: "deals at Databricks and Scale AI paid out employees and early investors without a listing."
Scale AI
- Description: AI data company.
- Why mentioned: Same liquidity-without-IPO example.
- Quotes: "deals at Databricks and Scale AI paid out employees and early investors without a listing."
- Description: AI lab.
- Why mentioned: Raised $11 billion, well short of its target, a sign of fundraising friction at the top of the market.
- Quotes: "OpenAI has secured $11 billion for its latest funding round, less than half of its $30 billion target, The Wall Street Journal reported."
- Description: Developer of an AI inference platform.
- Why mentioned: Raising at a $30 billion valuation, a signal of the inference-layer opportunity.
- Quotes: "Baseten, the developer of an AI inference platform, is raising a new round led by Sands Capital at a $30 billion valuation, Bloomberg reported."
- Description: Infrastructure for satellite networks.
- Why mentioned: Large Series C backed by top-tier firms.
- Quotes: "secured a $375 million Series C led by Andreessen Horowitz and DFJ Growth at a $2 billion valuation."
- Description: Munich-based European startup developing long-range hypersonic missiles.
- Why mentioned: Defense-tech deal in Europe.
- Quotes: "raised €100 million in a Series B led by Lightspeed and General Catalyst, valuing the Munich-based company at roughly €700 million, the Financial Times reported."
- Description: Robotics company.
- Why mentioned: Robotics Series B led by a16z.
- Quotes: "Robotics company Tutor Intelligence raised a $75 million Series B led by Andreessen Horowitz."
Specific Biologics
- Description: Biotech developing gene-editing therapies for genetic diseases.
- Why mentioned: Series A in gene editing.
- Quotes: "raised a $35 million Series A led by Hatteras Venture Partners."
Larridin
- Description: Developer of an AI value measurement platform.
- Why mentioned: Series A in AI ROI measurement.
- Quotes: "secured a $30 million Series A led by Pelion Ventures."
NPhase
- Description: Clinical data platform developer.
- Why mentioned: Late-stage clinical data raise.
- Quotes: "secured a $30 million Series D led by Viking Global Investors."
Pathfinder
- Description: Fintech startup that operates a platform for managing Education Savings Account funds.
- Why mentioned: Education fintech Series A.
- Quotes: "raised a $12 million Series A led by Portage Ventures."
Clark
- Description: Developer of AI employees for CPA companies.
- Why mentioned: Seed round in AI agents for accounting.
- Quotes: "raised an $11 million seed round led by Touring Capital."
Cloud Software Group
- Description: Vista- and Elliott-backed software company.
- Why mentioned: Debt refinancing.
- Quotes: "is offering senior secured notes to refinance $2 billion of debt."
OPay
- Description: SoftBank-backed Nigerian payments platform.
- Why mentioned: IPO halted by a regulatory inquiry.
- Quotes: "halted its IPO due to a regulatory inquiry."
Faire
- Description: Shopify-backed retail wholesale startup.
- Why mentioned: IPO candidate.
- Quotes: "is considering an IPO in the next year, Fortune reported. The company was last valued at $5.2 billion."
Mahle and Schott Pharma
- Description: German industrial companies.
- Why mentioned: Exploring IPOs of their Indian operations.
- Quotes: "both German industrial companies, are separately exploring IPOs of their Indian operations that could raise up to $400 million each, Bloomberg reported."
Omen AI and Aquatic Labs
- Description: Data center cooling-fluid monitoring tools companies.
- Why mentioned: Consolidation in a data center niche.
- Quotes: "Omen AI, which develops tools to monitor cooling fluids in data centers, agreed to acquire peer Aquatic Labs... in a $16 million deal."
Lovelytics
- Description: Data and AI services company.
- Why mentioned: Sponsor exit.
- Quotes: "Interlock Equity exited data and AI services company Lovelytics to CDW."
Lava
- Description: Adobe-backed customer engagement platform.
- Why mentioned: Acquired by Bilt.
- Quotes: "Housing rewards startup Bilt agreed to acquire Adobe-backed Lava, the developer of a customer engagement platform across sports, entertainment and retail businesses."
4. People Identified
Susan Hu
- Description: Quantitative Research Analyst at PitchBook.
- Why mentioned: Author of the IPO backlog analysis.
- Quotes: "By Susan Hu, Quantitative Research Analyst"
Jacob Robbins
- Description: Technology Reporter at PitchBook.
- Why mentioned: Author of the data center beneficiaries piece.
- Quotes: "By Jacob Robbins, Technology Reporter"
5. Operating Insights
Look for second-order beneficiaries of AI infrastructure demand
Operators and founders can find opportunity in the physical supply chain behind data centers rather than competing in crowded software categories.
"The AI boom's biggest beneficiaries may not write software at all." "channeling private capital into areas such as gas pipelines, fire suppression systems, and even cargo freighters."
Pure-play software must prove its value in an AI-driven market
Software companies can no longer rely on category tailwinds and need to demonstrate differentiated value.
"AI has put pure-play software companies under pressure to prove their value."
Plan for non-IPO liquidity paths for employees and early investors
Secondary and tender-style deals have become a viable alternative.
"deals at Databricks and Scale AI paid out employees and early investors without a listing."
6. Overlooked Insights
OpenAI's raise came in well under target
The fundraising shortfall is a potentially significant signal about capital availability, even at the top of the market, and it deserves more attention than a deal-roundup line.
"OpenAI has secured $11 billion for its latest funding round, less than half of its $30 billion target... The Financial Times reported that the company reduced the target funding amount in the round by $10 billion."
Real estate IRRs have deteriorated across successive vintages
This is a slow-moving structural trend relevant to LP allocation decisions.
"Median real estate IRRs have been declining since the global financial crisis rebound, with successive vintages getting worse from the mid-2010s to the early 2020s."