Investing in the Next Stars
H1 2026 Venture Secondary Market Report
H1 2026 · July 2026
Executive summary
Six months that felt like a decade. In the first half of 2026 we've cycled through tokenmaxxing and its hangover, watched a SaaSpocalypse narrative rise, waded through alleged AI-generated misinformation about U.S. datacenter build-out, sat through disaster-porn-as-marketing, collectively rebranded ourselves as SpaceTech investors — and now we're back to bubble worries as the BIS starts sounding warnings on hyperscaler capex.
Mid-2026 finds the market registering maximum fear and maximum greed at the same time. H1 2026 established a new benchmark for global venture investment, according to Crunchbase — but the headline number obscures a striking concentration: OpenAI and Anthropic alone attracted more than 40% of all venture funding in H1.
Even so, the broader venture ecosystem is showing genuine strength. Funding increased across every investment stage. Public markets have reopened. Billion-dollar rounds expanded beyond foundation model developers into AI infrastructure, defence, robotics, and healthcare. IPOs and startup acquisitions accelerated alongside venture investment, producing the strongest exit market since the 2021 boom.
This report is built on LBX.pro's proprietary secondary-market dataset — based on more than 25,000 listed bids and asks across H1 2026.
H1 2026 in the venture secondary market
Listed volume is up sharply — but slowing.
Listed interest volume in H1 2026 nearly doubled versus H2 2025. But Q2 2026 volume was 18% below Q1 — the initial surge is normalising, not compounding.
Demand mirrors the primary market's AI concentration.
Just as venture funding clustered around the AI race, secondary demand has clustered around the same theme: AI and AI infrastructure accounted for 41% of total bid volume in H1.
The market is looking for new stars — and the new stars aren't necessarily interested in secondary liquidity.
Demand has broadened well beyond the top names: the activity share of the LBX25 Index fell to just 56% in H1. A new cohort of bid-heavy companies is emerging, drawing significant buyer interest often without matched supply. As a result, the share of companies showing double-sided activity has dropped meaningfully.
Valuation growth on the secondary market decelerated in H1.
The LBX25 Index rose 28% in H1 2026, versus S&P 500 +8% over the same period. But drilling in, the median tells a different story: median growth for LBX25 companies was 22%, while for the ~300 other actively-traded names on the secondary market, median growth was 0%. In the mega-cap comparison, private mega-caps continue to substantially outperform public — median private Mag 7 return +33% vs. median public Mag 7 return +4%.
The broad market is trending toward deeper discounts — with AI Infrastructure and Mobility Tech as the clear exceptions.
In 2025, the reflex was to pay a premium for anything with a hot theme attached. In H1 2026 that reflex has reversed: buyers are now trying to acquire the same "new stars" at meaningful discounts, often unsuccessfully because supply doesn't materialise at those levels. The result is a market where individual quotes look almost random — aggressive bids sit unfilled, some tickets still trade above their last primary round, others trade 30–50% below. The re-pricing is real, but the path is uneven, and the pricing itself has become noticeably inconsistent.
This uncertainty mirrors what's happening in public markets. H1 2026 ended with a mixed picture for tech stocks — no clean directional signal, wide dispersion across names within the same theme, and repricing happening in different places at different speeds.
Three structural forces shaped the venture secondary market in H1
Hyperscaler competition is now full-stack.
The hyperscalers have moved from being infrastructure partners to being direct competitors at every layer of the AI stack — AI chips, training clusters, cloud AI rentals, developer APIs, frontier LLMs, enterprise AI applications, consumer AI applications, wearable AI devices, and humanoid robots. There is no longer a defensible category-of-one bet in AI that doesn't also compete against a hyperscaler.
Companies are actively restricting share transfers, and the retail synthetic instrument market has emerged as its own animal.
Anthropic, Anduril, and others tightened unauthorized secondary sale restrictions. In parallel, tokenized perpetuals and other retail-facing synthetic products emerged as a workaround — but the price signals from that market track sentiment rather than fundamentals, and the two markets have functionally decoupled.
The IPO window has re-opened at scale.
SpaceX went public on June 12 at a $1.77T valuation, closing 19% above the IPO price on debut and pulling $250B in investor demand — the largest tech IPO of the decade and the clearest signal yet that mega-cap private companies can now access public markets on their own terms. The pipeline behind SpaceX is remarkable: Anthropic, Lambda, Unitree, Agility Robotics and a lot more. For the first time since 2021, "wait for the IPO" is a credible liquidity path again.
Volume and activity share
Volume rose, but the growth isn't linear. Listed interest volume rose 95% in H1 2026 versus H2 2025 — but Q2 2026 was down 18% versus Q1, suggesting the H1 surge front-loaded into Q1 rather than building steadily through the half.
Breadth is expanding faster than depth. Active companies grew from 246 (H2 2025) to 352 (H1 2026), but a smaller share of that activity is genuinely two-sided — 70% of active companies were double-sided in H2 2025, versus just 58% in H1 2026. The shift is accelerating: Q2 alone saw bid-only and bid-heavy names (bid share >70%) rise to 74, from 48 in Q1. More names are trading, but more of that trading is one-sided price discovery rather than matched supply and demand.
Sector concentration remains stark. AI and SpaceTech (with SpaceX classified as SpaceTech) account for 32% of all listed volume. Narrow to the demand side alone and the concentration is even tighter — buyers are chasing these two sectors more aggressively than sellers are offering them.
Activity & bid share by sector
Activity share
Bid share
Source: LBXpro internal secondary-market data. Share of total H1 2026 activity and bid volume by sector.
The New Stars
A distinct new cohort emerged in Q2. 26% of companies active in Q2 had no listed activity at all in Q1 — one-sided demand showing up for names the market hasn't priced yet, well before two-sided activity develops.
| # | Company | Sector | Bid share | Valuation | Insights |
|---|---|---|---|---|---|
| 1 |  K2 Space | SpaceTech | 100% | $3.0B | GRAVITAS launched on Falcon 9 (Mar 2026) — first Mega Class satellite in orbit; $500M signed contracts. |
| 2 |  Fireworks AI | AI Infrastructure | 100% | $15B (target) | In talks to raise at $15B (3.75x its Oct 2025 round). ARR reached $800M in May 2026. Customers: Cursor, Perplexity, Notion, Uber. |
| 3 |  Chai Discovery | HealthTech / AI | 100% | $3.4B (target) | In talks to raise ~$400M at $3.4B — a 2.6x step-up from its $1.3B Series B six months prior. |
| 4 |  Modal Labs | AI Infrastructure | 100% | $4.65B | 4x jump from its Sep 2025 round. ARR grew $60M → $300M in 6 months; cloud partners 5 → 13. Serverless GPU infrastructure. |
| 5 |  FluidStack | CloudTech / Neocloud | 100% | $18B (target) | $50B data-center build with Anthropic; Google TPU deal; TeraWulf $6.7B hosting. Valuation $7.5B → $18B in 4 months. |
| 6 |  LangChain | AI Infrastructure | 100% | $1.25B | New Insights Agent + no-code agent builder. Customers: Replit, Clay, Harvey, Rippling, Cloudflare, Workday, Cisco. |
| 7 |  OpenRouter | AI Infrastructure | 100% | $1.3B | Universal API for 400+ LLMs from OpenAI, Anthropic and Google. |
| 8 |  Cognition | AI (Coding Agents) | 100% | $26B | ARR grew $37M → $492M (13x YoY). Customers: Goldman Sachs, Citi, Mercedes-Benz, Cisco, Palantir, US Army/Navy. |
| 9 |  Mach Industries | Defence | 100% | $1.8B | 4x valuation jump from its $470M Series B (Jun 2025). |
| 10 |  Field AI | Robotics | 100% | $2.0B | Certis partnership (Feb 2026) to deploy autonomous robotics in security operations. Nvidia-backed. |
| 11 |  Decart | AI (World Models) | 100% | $4.0B | 8x valuation step-up in 17 months. Amazon strategic customer + chip co-design partner. |
| 12 |  Wayve | Mobility Tech | 100% | $8.6B | London robotaxi trial with Uber launches 2026; consumer vehicles from 2027. |
| 13 |  Project Prometheus | Artificial Intelligence | 66% | $41B | Bezos-backed 'artificial general engineer' for design → manufacturing. Series B closed $12B at $41B. |
| 14 |  Starcloud | SpaceTech | 40% | $2.2B (target) | First pure-play orbital data-center unicorn. Raising ~$200M at ~$2.2B — roughly double its Series A a month earlier. |
| 15 |  LeoLabs | SpaceTech | 25% | n/d | US Space Force contract (Dec 2025); NASA conjunction-assessment participation. LEO tracking-radar network. |
| 16 |  Positron | AI Hardware | 23% | $1.06B | Custom AI inference hardware; positioned against Groq and Cerebras. Rapid recent momentum. |
Source: LBXpro internal secondary-market data. Bid share reflects secondary-market demand intensity over H1 2026. Valuation = latest round or reported target.
New stars: valuation velocity
Of the 16 companies we track as New Stars, ten became unicorns in just the last six months. Valuation velocity across the cohort is unprecedented — five names stand out:
| Company | Step-up | Window | Valuation move |
|---|---|---|---|
| Decart | 8x | 17 months | $500M → $4B |
| Fireworks AI | 3.75x | 7 months | $4B → $15B |
| Cognition | 2.5x | 8 months | $10.2B → $26B |
| Modal Labs | 4x | 8 months | $1.1B → $4.65B |
| FluidStack | 2.4x | 4 months | $7.5B → $18B |
The infrastructure/inference layer is where concentration is most extreme. Six of the sixteen New Stars sit in AI Infrastructure/Neocloud/Inference, and within that group valuations compress on unusually short timelines — Modal (4x in 8 months), Fireworks (3.75x in 7 months) and Positron (~5x in 5 months) are priced less like software companies and more like scarce infrastructure.
Strategic investor–customer overlap is now the norm. Nvidia, Amazon, Google and Anthropic each appear as both a customer and a cap-table participant across multiple names — Decart, Wayve, FluidStack and Fireworks all show this dual role.
Secondary premiums / discounts
Secondary pricing has moved decisively toward discounts. The market-wide median discount widened to −40% as of 2 July 2026, from −31% on 1 January 2026 — a six-month deterioration of roughly 9 points.
The pattern holds even within our curated LBX25 basket, though from a much stronger starting point: median pricing there sits at −8% today, down from +5% on 1 January 2026 — a swing from premium to discount over the half.
Deepening discounts are closely tied to activity structure. Companies with genuine two-sided activity carry a median of just −9.5%, versus the much wider market-wide −40% — confirming that the bid-only and bid-heavy names are dragging the broader discount lower. Where real two-way price discovery exists, pricing is holding up far better than where it doesn't.
The divergence is stark. AI Infrastructure, SpaceTech, Defence and Mobility all trade at premiums and are mostly widening — the market is still willing to pay up for compute, space, defense tech and autonomy. Everything else trades at a discount and is getting worse: SaaS and Crypto now both around −69%, and HealthTech deteriorating fastest (−18 points in six months). Robotics and CloudTech/DevOps flipped from premium to discount over the period.
Source: LBXpro internal secondary-market data. Median premium/discount to last primary round; current = 2 Jul 2026, prior = 1 Jan 2026.
Returns
The gap between public and private mega-caps is striking. Even the strongest public name (Apple, +21.5%) trails every private Mag 7 name except OpenAI. Anthropic's +135% dwarfs the entire public cohort by an order of magnitude.
Public Mag 7 · 6-month growth
Apple+21.5%
Nvidia+9.5%
Alphabet+4%
Meta Platforms+4%
Amazon+2%
Tesla−12%
Microsoft−17%
Median 6-month growth among active companies
The LBX25 Index is up +22% over the half, against a flat 0% median for all other active companies — as always, the most active companies are meaningfully outperforming the broader market.
AI Infrastructure is the standout at +49%, more than double the next-best sector — consistent with the premium and activity data elsewhere in this report. Crypto and SaaS are the clear laggards, both firmly negative, echoing the discount widening seen in the pricing section.
H1 2026 growth champions
Beyond the sector medians, H1 2026 produced our favourite outlier cohort — individual names posting extraordinary six-month growth well outside their sector averages.
| # | Company | Sector | 6-month growth |
|---|---|---|---|
| 1 |  Unitree | Robotics | +1,367% |
| 2 |  Booster | SpaceTech | +400% |
| 3 |  ClickHouse | CloudTech & DevOps | +198% |
| 4 |  Neuralink | HealthTech | +187% |
| 5 |  Skild AI | Robotics | +175% |
| 6 |  SambaNova Systems | AI infrastructure | +168% |
| 7 |  Replit | CloudTech & DevOps | +152% |
| 8 |  Lovable | Artificial Intelligence | +143% |
| 9 |  Temporal | CloudTech & DevOps | +129% |
| 10 |  OpenEvidence | HealthTech | +125% |
| 11 |  The Boring Company | Mobility Tech | +125% |
| 12 |  Ayar Labs | Artificial Intelligence | +121% |
| 13 |  Hadrian | Robotics and Drones | +117% |
| 14 |  Motive | Mobility Tech | +110% |
| 15 |  d-Matrix | Artificial Intelligence | +107% |
Source: LBXpro internal secondary-market data. Change in implied valuation over H1 2026 (Jan–Jun).
Cooling market, diverging bets
Cooling shows up in the numbers underneath every section of this report. Q2 2026 volume is down 18% versus Q1. Pricing tells the same story: the market-wide median discount widened from −31% to −40% over the half, and even the LBX25 basket flipped from a +5% premium to an −8% discount.
Against that backdrop, investors are making bets.
Some are buying stability.
Databricks and Stripe are the closest things the private market has to blue chips: large, profitable-adjacent, and structurally less exposed to a single AI-demand shock than almost anything else on this list. Their growth is real but unspectacular by 2026 standards — which is precisely the point. This is capital choosing to sit out the most volatile part of the market without leaving private markets altogether.
Some are rotating defensive.
Defence, SpaceTech and Mobility are holding premiums and posting solid, unspectacular growth — a bet on sectors where AI is a tailwind rather than the entire thesis. Notably, these are also the sectors best insulated from the broader discount-widening described above.
And some are chasing outliers.
This is capital that has explicitly decided the payoff from being early in AI, AI Infrastructure and Robotics outweighs the risk of being wrong about the exact multiple.
Bubble or not, this is what the private market has always done at moments of genuine technological uncertainty: split into those preserving optionality, those hedging into adjacent stability, and those betting the whole thesis on being early.
Source: LBXpro internal secondary-market data · LBX Pro Research · July 2026