LIGHT HOPE

x1000.ai

SunVax mRNA Therapeutics

sunvaxmrna.com

LIGHT HOPE Quarterly · Issue 1 · 2026 Q2

Strategic Overview战略思路总览 · Built for SunVax · With Respect

Published 2026-05-22 · 15-min read · Audit-graded · NCNU 5-禁 lint PASS

§0 · Before anything else — a note on respect首先 — 一段关于尊重的话

This page exists because of a specific exchange on 2026-05-15 at the Antibody Conference panel. Yingzhong Li, founder of SunVax, was asked whether his lipid-nanoparticle discovery process leans on machine learning. His answer was direct and unhedged: empirical screening is the real workflow today, and ML-designed LNP is a five-year horizon at best. That kind of on-the-record honesty from a founder is rare.

中文: 本页源于 2026-05-15 Antibody 大会 panel 一次具体的对话 — 李英中先生在被问及 LNP 发现流程是否倚赖机器学习时,他的回答直接而坦诚:经验筛选才是真实工作流程, ML 设计 LNP 还要五年。这种创始人级别的坦率,不多见。

What follows is the result of reading every public SunVax artifact we could find — the company website, the 14 SV-Delivery™ kits, the panel transcript, Yingzhong's peer-reviewed papers (PMID 40851032 in Communications Biology 2025 on CATP; PMID 31061426 in Scientific Reports 2019 with the Irvine and Weiss labs at MIT Koch Institute) — and asking three questions in good faith. We have not asked SunVax for any internal data. Everything below is sourced from public material SunVax has already published, and every claim is point-clickable to a primary source.

The discipline that anchors this work: we never frame SunVax as "AI-driven." Yingzhong stated his methodology publicly; we respect it. Our linter blocks the contradictory framing at generation time, on every file, before any publication.

§1 · Question 1 — What does AI do for SunVax?问题一 — AI 对 SunVax 能赋能什么?

Short answer (一句话):

AI does not design SunVax's lipids. AI does the slower, more disciplined work of auditing everyone else's claims about SunVax's field — with verified_at timestamps the customer can re-check.

AI 不做 SunVax 的脂质设计。AI 做的是更慢、更有纪律的事:对外部对 SunVax 所在领域的全部 claim 进行审计 — 每条 claim 都有 verified_at 时间戳,客户随时可以重新校验。

What we offer is the slower work of mapping the market landscape SunVax operates in — competitor positioning, modality lines, where the field is moving in the next 18-36 months. Every source we cite resolves to a live URL. Every claim carries an evidence grade. Every quarterly issue carries a refresh log. The point is not that we audit anything you wrote; the point is that you can audit what we wrote, in 30 seconds, by clicking through.

§2 · Question 2 — What is SunVax's defensible position?问题二 — SunVax 的可立足位置在哪?

Short answer (一句话):

14 cell-type-specific kits · no antibody conjugation · empirical screening grounded in a proprietary ionizable lipid library. The exact place where the next decade of EGFR+ NSCLC immunotherapy needs delivery infrastructure.

14 个细胞类型特异 kit · 无抗体偶联 · 经验筛选 · 自有 ionizable lipid 库。 EGFR+ NSCLC 免疫治疗下一个十年所需要的 delivery 基础设施 — 正好就是这个位置。

EGFR-mutated NSCLC is the paradoxical center of solid-tumor oncology: five generations of TKIs, four bispecifics, and one ADC are approved (≈$15B/year US market), yet 5-year survival remains under 20% because the immune microenvironment is "cold" (low TMB, M2 TAM-dominated, ≈63% HLA LOH in L858R cases per Lin et al PMC9727402).

Three emerging immunotherapy modalities each need cell-type-tropic LNP delivery — and the SV-Delivery™ portfolio maps directly to each one:

Modality 1

Neoantigen mRNA vaccine

Mittal lab (Weill Cornell) · Abogen ABO2013 Phase 1 · Moderna V940 · BioNTech BNT116

→ SV112 lymphatic + SV110 mono/mac

Modality 2

mRNA-encoded EGFR×CD3 bispecific

BioNTech LiTE-RNA · Promab HER2-CD3 analog · multiple preclinical

→ SV110 + SV106

Modality 3

In-vivo CAR-T LNP enhancement

Mittal lab EGFR L858R TCR-T · EsoBiotec ESO-T01 BCMA precedent

→ SV102 / SV106 / SV108

A separate ten-company competitive landscape (Capstan, Acuitas, Genevant, Orna, Strand, Nutcracker, Generate Biomedicines, EsoBiotec/AZ, Mittal lab, Abogen) — scored on threat × overlap × posture — is delivered to SunVax leadership under NDA. The summary: SunVax's IP sits in the lipid library (not derivative of antibody-conjugation patents); evidence is empirical (the path regulatory and payer environments currently prefer); the portfolio is broad (not paradigm-locked to a single modality).

§3 · The five pain points we read in your work我们从你的工作中读到的五个痛点

Reading the public catalog, the conference slides, and the panel transcript, five concrete operational pain points surface. We do not claim to solve them for you. We do offer infrastructure that makes four of them more tractable in the conversations SunVax has with R&D customers and pharma BD partners. The fifth is the founder's domain, and we respect that boundary.

Pain 1 · Kit data scattered across slides, Excel, lab notebooks

Addressed

14 kit 的实验数据分散 · 客户教育成本高

What we offer: 13 per-kit pages on this site, each datum sourced and evidence-graded; kit-selector returns top-3 matches in <5 minutes; structured replacement for sales email cycles.

Pain 2 · 1-3 week sales-email cycle for kit selection

Addressed

客户买 kit 要 1-3 周决策周期

What we offer: kit-selector · experiment-designer · troubleshooter — auto-routes to order.sunvaxmrna.com when buyer is ready.

Pain 3 · No standardized data-room for pharma partnership

Available when needed

缺乏标准化 pharma data room

What we offer: NDA-gated partner-room subdomain template ready to provision when SunVax has a specific BD conversation. Not pushed today — available on request.

Pain 4 · CMC-readiness gap when customers approach IND

Available when needed

客户进 IND 时缺 CMC 准备度支持

What we offer: pre-IND scale-up readiness audit framework, customized to cell-type-specific LNP. Not pushed today — available on request.

Pain 5 · Chemical space 10¹⁰ vs data 10³–10⁴

Founder's domain — out of scope

化学空间 vs 数据量的 scale 问题

What we will not offer: ML-designed LNP, computational lipid discovery, or any framing that contradicts the founder's stated empirical methodology. This pain point belongs to the wet lab; the infrastructure we offer does not extend into it.

§4 · The LIGHT HOPE thesis — in one diagramLIGHT HOPE 的核心主张 · 一图概览

Thesis

Our infrastructure makes SunVax's empirical position more defensible, not less.

我们的基础设施让 SunVax 的经验主义立场 更可立足,而不是相反。

Q1 · AI赋能

Audit pipeline · 5-stage workflow · verified_at timestamps

Q2 · SunVax赋能

14 kits · empirical · proprietary lipid library · 3 modality fits

Q3 · 合作模型

Bilateral · revocable · 1-business-day takedown · respect-first

§5 · The discipline (what we will not do)纪律 — 我们不会做的事

✗Will not market SunVax with methodology labelling that contradicts the founder's stated position. 不会以与创始人公开立场矛盾的方式标识方法论。
✗Will not produce content that crosses into clinical advice, treatment overpromise, or patient-specific forecasting (NCNU 5-禁 P_1 / P_2 / P_3).
✗Will not claim regulatory endorsements that do not exist (P_4).
✗Will not publish anything SunVax asks us to remove — 1 business day takedown SLA. 1 工作日撤稿 SLA。
✗Will not forward SunVax intel to other customers — confidentiality protocol §7.
✗Will not push commercial conversations SunVax has not asked for. 不主动推 SunVax 没有提出的商业对话。

§6 · Further reading延伸阅读

/industry-map — a sharper second version of this page, aligned to the three questions Yingzhong asked at the panel (AI's real role in mRNA delivery · SunVax's market coordinates · the R&D kit vs pharma partnership tradeoff). Newer than this page; recommended over this one if you only have time for one.
The 30-page strategic brief — delivered separately under NDA-INTERNAL data class. Every claim source-linked + evidence-graded + verified_at timestamped.
/insights/how-this-site-is-built — the discipline behind every claim on this site
/insights/empirical-vs-ml-lnp — Yingzhong's on-the-record empirical methodology
/about — methodology + takedown commitment
/contact — to correct, ask, redact, or end the engagement