Lux Capital

NYC-based frontier-tech and AI applications venture firm.

Snapshot

• Founded: 2000 by Peter Hébert and Josh Wolfe
• HQ: NYC
• AUM: ~$7Btotal(post-LuxVenturesIX,Jan2026;44people).[LuxVenturesIXclosed$1.5B in Jan 2026](https://www.luxcapital.com/) — the firm’s largest fund ever
• Stage: Multi-stage
• Original thesis: Frontier deep-tech (defense, biotech, semis, robotics)
• 2024-2026 expansion: AI applications via Grace Isford and Brandon Reeves (software side)
• Cap-table conflict status: Verified clean across all 8 Init Intelligence competitors as of 2026-05-05.

Why this is a clean lane

Lux is actively in AI applications but flavors itself toward frontier-feeling bets (foundation models, AI-for-engineers, biology, robotics). Their AI services exposure is Maven AGI (customer support agent) — adjacent enough to be thesis-positive, far enough not to conflict with AI-ITSM. No portfolio conflict.

Lead deal partners

Partner	Focus / thesis	Portfolio anchor
Grace Isford	Partner; “AI/ML infrastructure stack”	Maven AGI, LangChain, Runway, Sakana, Hugging Face
Lan Jiang	The actual warm front door (per the 5/5 meeting). Sourcing engine for both relevant in-window seeds — sourced Applied Compute + championed Tenzai. Title ^[ambiguous] (firm site “Partner”; some trackers “Associate”).	Applied Compute, Tenzai
Brandon Reeves	Software/AI/defense; sits across most of the AI app investments	Cognition (Devin), Applied Compute, Hugging Face, Benchling
Bilal Zuberi	Co-lead on Maven AGI; broader applied-AI lens	Maven AGI

Relevant portfolio (AI apps + agent infra)

• Maven AGI — Seed/A $28Mco-ledwithM13(Mar2024);SeriesB$50M (Jun 2025, Dell Tech Cap led) — Grace Isford + Bilal Zuberi — AI customer support agent; closest analog to AI ITSM
• Cognition (Devin) — $400MSeriesBat$10.2B (Sep 2025) — Brandon Reeves — AI software-engineer agent; services-as-software pattern
• Applied Compute — Seed/A — Brandon Reeves + Lan Jiang — “in-house agent workforce” for enterprises
• Hugging Face — Multi-round — Brandon Reeves — foundational AI infra
• LangChain — Series A — Grace Isford — agent orchestration
• Together.xyz — Multi-round — open model infra
• Modal — Multi-round — Erik Bernhardsson + Akshat Bubna — serverless agent compute
• Runway — Series C+ — Grace Isford — generative video
• Sakana AI — Multi-round — foundation models

Related

• cap-table-patterns-across-startup-competitors
• Init Intelligence

---

Fundraise intel (2026-05-25)

Firm Profile (reconciled + updated)

• Founded: 2000 (NYC) by Peter Hébert and Josh Wolfe. Now NYC + Silicon Valley.
• Thesis: “Turn science fiction into science fact” — emerging science & technology “at the outermost edges of what’s possible,” backing scientists/engineers/polymaths over finance-first investors. One of the earliest institutional dual-use/defense backers (“reintroduce the Pentagon to Silicon Valley”). Operates an in-house media platform. Lux Labs has created/co-founded 20+ de novo companies.
• Fund / AUM — ⚠️ correction vs the content above: the page above says ”~$5B total AUM.” That was the post-Lux-8 figure. Updated, double-verified:
  • Lux 8 (8th core fund) = $1.15B\ast \ast ,raised\ast \ast 2023\ast \ast \to broughtAUM\ast \ast over$5B.
  • Lux Ventures IX (Fund IX) = $1.5B\ast \ast ,closed\ast \ast Jan2026\ast \ast (oversubscribed;fir{m}^{\prime }slargestever)\to \ast \ast totalAUMnowover$7B. (The page above correctly states the $1.5BJan-2026IXnumber;onlythe"$5B total” is stale — it’s now ~$7B.) ^[verified: Sidley, TechCrunch, Bloomberg, VCWire, Lux blog, Wikipedia]
  • Team: ~44 investment professionals manage the $7B.
• Stage / check: any-stage; check sizes $100K\to$100M (“first $100Kandthelast$100M”). Runs a single large multistage vehicle (not a separate seed fund) → partners can write/champion early checks; exact unilateral IC authority not public ^[inferred].

Partners (init.inc-relevant — pages now exist)

• Lan Jiang — generalist (biotech/healthcare-computational + ML infra + dev tooling + applied AI). init.inc’s primary & warmest Lux relationship (met 5/5, intro’ing to Grace). Sourced Applied Compute; championed Tenzai. (Title note: Lux site + CRM say “Partner”; external trackers say “Associate” — ^[ambiguous].)
• Grace Isford — Partner (youngest in Lux history), lead AI/ML-infra + LLM-apps voice; Forbes 30U30. The target decision-maker. Led LangChain, Runway, Sakana, Maven AGI, Phonic.
• Brandon Reeves — GP (software/AI/defense/manufacturing). Led Anduril, Hugging Face, Anchorage, Chronosphere, Benchling, Cognition. Secondary/backup contact (via Kushal).
• (Bilal Zuberi — co-lead on Maven AGI; not init.inc-targeted this pass.)

Recent seed activity (Pass-1 sketch — ≤5 deals; exhaustive 18–24mo pull is Pass 2)

• Tenzai — $75M seed, Nov 2025 (co-led Greylock + Battery + Lux). Agentic AI pentesting; Brandon Reeves named, Lan Jiang championed. One of the largest cybersecurity seeds ever.
• Applied Compute — $80Mseed,Oct2025(Benchmark+Sequoia+Lux+angels;$700M val ^[single-source]). Ex-OpenAI; “specific intelligence” in-house agent workforce. ⚠️ thesis-adjacent to init.inc, different buyer. Lan Jiang sourced; Brandon Reeves linked.
• Phonic — $4M seed led by Lux, Apr 2025. Voice AI (MIT founders). Grace Isford’s deal.
• Together AI — $305MSeriesB(Feb2026,$3.3B val) — Lux participated as an early backer (not a seed; context for Grace’s AI-infra anchor).
• Maven AGI — $50M Series B (Jun 2025, Dell Tech Cap led) — Grace + Bilal Zuberi; carried from the table above (closest Lux analog to AI-ITSM).
• Full data + verification → /Users/sazzad14/seed-market-scan/firms/lux-capital.csv.

Best data channels for this firm

Lux’s own site is unusually rich — luxcapital.com/news carries quarterly LP letters (Josh Wolfe’s are detailed, name specific deals) and named investment memos, plus full partner bios (the partner bios are the source of truth for who-works-with-what). Grace Isford’s own essays (luxcapital.com/news + graceisford.com) are the best thesis signal. For deal amounts/leads: TechCrunch + BusinessWire + regional press (Tenzai = Israeli/cyber press). Aggregators (vcsheet/NFX/RocketReach) are good for partner profiles but conflict on Lan Jiang’s title (Associate vs Partner) — trust the firm site + the live meeting.

Fundraise Relationship (init.inc)

• Relationship facts: Lux is a clean, thesis-positive lane (no portfolio conflict across init’s 8 competitors — confirmed on the page above). Live path: Lan Jiang (front door, met 5/5) → internal intro to Grace Isford (decision-maker). This internal path supersedes the stale external Grace path (Aaron Lee messaged Grace 4/28, no reply). Lux’s two thesis-adjacent bets have different buyers: Maven AGI (CX) and Applied Compute (model-training infra), vs init’s IT-ops buyer. Open question (from the page above): whether Grace reads back-office IT as “frontier enough.”
• Conversation log (newest first):
  • 2026-05-05 14:30 SF: Sazzad met Lan Jiang — went really well; Lan intro’ing to Grace internally. (Strongest Lux path.)
  • 2026-04-28: Aaron Lee → Grace external message, no reply (superseded).
  • 2026-04-25: Arjun intro’d Lan.
• Mutual connections / warm path: Arjun (→ Lan, primary), Sheryl, Kushal (backup, multi-partner), Aaron Lee (→ Grace, stale).
• Live stage: tracked in the CRM (Lan = Follow Up Phase; Grace = Intro Made; Brandon = Secondary).

Cross-check vs existing wiki

• The “clean lane / no conflict” framing above is reaffirmed. The only correction is the AUM (~$5B\to$7B post-Lux IX) and the Lux 8 vs Lux IX disambiguation (both now stated). Partner targeting (Grace primary, Brandon secondary) is consistent; Lan Jiang is newly elevated as the actual warm front door per the 5/5 meeting. No ⚠️ contradiction beyond the AUM staleness.

---

Pass 2 deal pattern (2026-05-25)

Exhaustive ~18-24mo (Nov 2024 → May 2026) pre-seed/seed pull, triangulated ≥2 sources/row. Full 12-row table → /Users/sazzad14/seed-market-scan/firms/lux-capital.csv. This block is additive; nothing above is edited.

Seed lead-rate & deal counts (in-window)

Lux is a multistage deep-tech firm that announces very few pure seeds — exactly the playbook §5.8 pattern (a big single multistage vehicle, not a dedicated seed fund). Across the full window, confirmed in-window announced seeds = 3-4: Phonic ($4M,LuxLED),\ast \ast AppliedCompute\ast \ast ($80M, co-led), Tenzai ($75M, co-led), plus Lumaril (2025, Lux-led, amount undisclosed). Tracxn shows ~31 new Lux investments in the trailing 12mo — most are unannounced or later-stage; the announced seed surface is small by design. So Lux’s seed lead-rate among announced seeds is high (it led/co-led every announced in-window seed it appeared in), but the volume is low vs a dedicated-seed firm like Felicis.

Check / stage profile

• Lux genuinely deploys “first $100K\to last$100M” (confirmed firm language) — so check size is wildly bimodal.
• The one small classic seed (Phonic $4M,Luxsolo-lead)sitsalongsideitshistoricalbiotech/AIseedrange($4-5M, e.g. Inductive Bio $4.3M co-led, Dec 2023, out-of-window).
• The big in-window “seeds” (Applied Compute $80M,Tenzai$75M) are frontier/AI-talent mega-seeds where Lux co-leads alongside Benchmark/Sequoia/Greylock/Battery — not solo bets.
• Lux leads Series A in deep-tech readily (Reflect Orbital $20MSeriesALED;Atrandi$25M Series A LED) — its lead muscle shows more at A than at seed for capital-intensive science.

Cadence

Low announced-seed cadence — a handful of disclosed seeds across 18 months, irregular. Most activity is invisible (Lumaril-style quiet checks, follow-ons, later-stage). Don’t over-read a near-empty announced-seed set as inactivity — it’s a multistage firm’s normal recall gap (two independent sweeps agree on the short list).

Sector mix

Frontier deep-tech: AI infra / agents (Phonic voice agents, Applied Compute agent-workforce, Together AI infra), cybersecurity (Tenzai agentic pentesting), space / defense (Reflect Orbital, Lux Aeterna — not a Lux deal, see below), biotech/life-sci tools (Atrandi, Inductive Bio), AI hardware (Majestic Labs memory wall), industrial AI (Lumaril). The agentic-AI cluster (Phonic, Applied Compute, Tenzai, Maven AGI) is the init.inc-adjacent neighborhood — but each has a different buyer than IT-ops.

Partner attribution

• Grace Isford (Partner; target decision-maker): Phonic (LED + quoted, confirmed in TechCrunch + PRNewswire). Anchors AI-infra (Together, LangChain, Runway, Sakana, Maven AGI).
• Brandon Reeves (GP; secondary contact): Tenzai (named on deal), Applied Compute (linked). vcsheet pegs his style as seed-lead, $2M–$20M checks.
• Lan Jiang (init.inc’s warm front door): sourced Applied Compute (Stanford ties) and championed Tenzai. (⚠️ title: Lux site = “Partner”; some trackers = “Associate” — ^[ambiguous]; Lan Jiang should be added to the Lux partner table per coordinator note.)
• Bilal Zuberi / Deena Shakir — Maven AGI (CX) / bio respectively; not init.inc-targeted.

Notable seeds (init.inc-relevant)

• Phonic ($4M, Grace Isford LED, Apr 2025) — proves Grace writes small lead checks into reliability-focused agent startups; the cleanest precedent for an init.inc Grace pitch.
• Tenzai (75M co-led, Nov 2025) & **Applied Compute** (80M co-led, Oct 2025) — show Lux co-leading mega-seeds in agentic AI / security, with Lan Jiang as the sourcing engine (the warm path).

Best data channels (refined)

Lux’s own site is the spine but it’s thematic, not a deal table. luxcapital.com/companies/<name> pages carry a “Lux Investment: YYYY” milestone (how Lumaril/Reflect Orbital/Majestic were caught) — but no amount/role, and the index is hard to enumerate. The quarterly LP letters (/news, Josh Wolfe) name select deals in prose, not tables. For amounts/leads: TechCrunch + BusinessWire/PRNewswire + regional press (Tenzai/Atrandi = Israeli/Lithuanian press; Calcalist/CTech). Aggregators (vcsheet, Tracxn, CB Insights, NFX Signal) give partner profiles + portfolio counts but are sparse on dated seed rows and conflict on Lan Jiang’s title. Net: Lux requires per-company press triangulation, not a clean firm-level pull.

⚠️ Pass-2 corrections & disambiguations (existing text left unedited)

• Maven AGI: existing wiki + Pass-1 CSV called it a “Seed/A $28M"anda"$50M Series B.” Primary sources → the $28Mtotal=a$20M SERIES A LED BY M13 (May 2024), Lux participated (NOT a Lux seed/lead). It’s also out of the Nov-2024 window. The later $50M (Jun 2025, Dell Tech Cap) is the Series B. Corrected in CSV.
• Applied Compute valuation: Pass-1 said ~$700M(single-source).Refined\to \ast \ast$500M** (SiliconANGLE); $700M was a during-raise figure.
• Lux Aeterna ≠ Lux Capital (name collision): the $4M pre-seed reusable-satellite startup “Lux Aeterna” (Jun 2025) was led by Space Capital; Lux Capital is NOT an investor. Excluded.
• AUM / team (deferred frontmatter facts — VERIFIED): Lux manages **~$7Bwith44people\ast \ast post-LuxVenturesIX($1.5B, Jan 7 2026) — confirmed via the Lux IX announcement post and TechCrunch (“valued at about $7billioninAUM").Thefrontmatterstillsays"$5B total AUM” → should be updated to ~$7B (coordinator action).
• Lan Jiang partner table: per coordinator note, Lan Jiang should appear in the Lux partner table (she sources init.inc’s two most relevant in-window seeds). Title remains ^[ambiguous] (Partner per firm site).