Eliyan
Eliyan:面向 AI 数据中心的 UCIe Chiplet 互连
Eliyan 的战略相关性强,技术差异化也清楚,但公开可见的独角兽估值仍跑在可见经营证明前面。
封面要素
公司概况
Eliyan 是一家位于 Santa Clara 的半导体基础设施创业公司,为 AI、云和高性能计算系统开发 chiplet 互连技术。公司的产品叙事围绕高带宽裸片到裸片、存储和新兴电光连接展开,已获得主要计算、平台和基础设施参与者的生态验证,但公开经营披露仍然有限。
- 成立时间
- 2021-01-01
- 创始人
- Ramin Farjadrad
- 创立地点
- Santa Clara, CA
- 总部
- Santa Clara, CA
- 产品
- 面向多裸片 AI 系统的连接 IP、PHY、chiplet 和存储互连技术,路线图正延伸到电光互连。
- 客户
- 设计先进 AI 系统的 hyperscaler、AI 加速器厂商、存储供应商和基础设施平台团队。
- 商业模式
- 通过 design-in 项目销售深层基础设施半导体连接技术,可能覆盖 IP、PHY、chiplet 及相关工程 / 商业化合作。
- 阶段
- Series C
- 融资情况
- $145M Series C,估值 $1B(2026 年 7 月)
执行摘要
主要优势
- AI 连接瓶颈真实存在,Eliyan 的产品能切入 die-to-die、内存以及靠近光互连的架构。
- 计算、网络和光学领域的战略方与合作伙伴背书,给了生态验证。
- 2026 年大额融资给公司争取了商业化和路线图里程碑的时间。
主要风险
- 公开收入、利润率、现金跑道和 backlog 披露仍太少,难以按常规方式承保。
- 客户留存韧性和集中度仍靠推断,尚未由具名量产证明打实。
- 资质验证、封装和出口管制复杂度都可能拖慢商业转化,执行风险偏高。
未决问题
- 需要客户层面证据,证明评估项目正在转成可重复的生产部署。
- 需要收入结构、毛利率、客户集中度和现金跑道数据,检验 $1B 估值是否合理还是偏撑。
目录
01公司概览
1.1 身份与产品范围
Eliyan 将自己定位为一家先进连接半导体公司,瞄准越来越定义 AI 基础设施的带宽、存储和功耗瓶颈。从官网、产品页、技术页到 2026 年 7 月 Series C 公告,公司始终把核心产品描述为一组 NuLink PHY IP 模块和 NuGear chiplet 产品,覆盖裸片到裸片、裸片到存储、芯片到芯片,并逐步延伸到机架到机架场景。故事里最差异化的地方不只是原始速度,而是封装灵活性:Eliyan 主张,其互连能在标准有机基板上交付接近先进封装的性能,从而降低成本、复杂度和供应链依赖。这个点重要,因为 AI 系统建设者正撞上中介层容量、热密度、测试复杂度和上市周期的现实上限。因此,Eliyan 卖的不是一小块狭义 IP,而是一套连接架构判断,并把自己定位为把 chiplet 经济性延展到计算、存储和光互连织构的路径。[CO001, CO003, CO004, CO005, CO006, CO007]
| 指标 | 数值 / 状态 | 日期 | 置信度 | 缺口 |
|---|---|---|---|---|
| 总部 | 加利福尼亚州 Santa Clara | 2026-07-29 | 高 | None |
| 成立时间 | 2021 | 2025-11-14 | 中 | 未找到官方成立公告;来源为第三方数据库 |
| 当前阶段 | Series C 轮 | 2026-07-29 | 高 | None |
| 核心产品范围 | 面向 D2D、D2M、C2C 和近光互连连接的 NuLink PHY IP 以及 NuGear chiplet 系列 | 2026-07-29 | 高 | 客户维度的具体产品组合未披露 |
| 最新轮次 | $145M Series C 轮,估值 $1B | 2026-07-29 | 高 | 私有股权结构未披露 |
| 2022 年以来公开披露融资额 | 至少 $295M,覆盖 Series A、Series B、2026 年 1 月战略融资和 Series C | 2026-07-29 | 中 | 基于公开轮次算术相加;公司未公布累计融资总额 |
| 已披露的 Series C 新战略投资方 | Cisco Investments;Lumentum | 2026-07-29 | 高 | 其他参投方未完整列出 |
| 核心领导层 | Ramin Farjadrad(CEO)等核心领导:Patrick Soheili、Syrus Ziai | 2026-07-30 | 高 | 更完整组织架构未完全披露 |
| 客户进展状态 | 公司称,领先的 hyperscale、AI 加速器、内存和基础设施客户正在评估或部署 Eliyan 技术 | 2026-07-29 | 中 | 已审阅材料中没有具名量产客户 |
| 员工数 | Tracxn 数据显示,截至 2026 年 6 月有 97 名员工 | 2025-11-14 | 低 | 第三方估计;公司未公布官方人数 |
| 营收 / ARR | 未公开披露 | 2026-07-30 | 中 | 需要管理层披露或一手财务记录 |
| 治理机制 | 董事会新增成员已公开;所有权、保护性条款和委员会细节未披露 | 2026-07-30 | 中 | 需要章程、融资文件或管理层披露 |
基于官方公告、产品页和第三方公司数据库拼出的时点型公司概览。各行有意把缺少支持的私营公司指标保留为缺口,而不是自行推算。
[CO001, CO002, CO003, CO009, CO010, CO023]Eliyan 的商业逻辑把 chiplet PHY IP、chiplet 产品、战略投资人和 AI 系统瓶颈连在一起。
[CO003, CO004, CO008, CO022, CO023, CO025]当前公开图景:融资和路线图里程碑强,披露经营指标弱。
[CO009, CO010, CO023, CO035, CO040]1.2 创始人、领导层与治理
公司仍由创始团队主导,这也是最重要的尽调事实之一,因为 Eliyan 的技术可信度和 go-to-market 执行,都押在少数几位连续创业 / 运营者身上。官方材料显示,Ramin Farjadrad 是联合创始人兼 CEO,Patrick Soheili 是联合创始人并负责当前业务战略,Syrus Ziai 是联合创始人并承担运营或工程领导角色。创始人履历与 chiplet 互连创业公司的市场需求高度贴合:Farjadrad 早前共同创办 Velio 和 Aquantia,并参与过后来进入 BoW、再影响 UCIe 的标准工作;Soheili 有高管、创投和半导体 go-to-market 经验;Ziai 则在多家先进硅团队担任过工程领导。治理已扩展,但仍偏私营公司轻量披露。公开证据确认 Jason Taylor 于 2023 年加入董事会,Umesh Padval 于 2026 年加入;公开团队页还显示公司在财务、法务、销售、存储产品和接口 chiplet 上有更深的人才梯队。仍然缺失的是委员会结构、所有权比例、投票权,以及晚期投资人通常需要直接审阅的其他控制机制。[CO012, CO013, CO014, CO015, CO016, CO017]
| 人物 | 角色 | 背景 | 创始人-市场匹配 / 职能覆盖 | 关键人物依赖 |
|---|---|---|---|---|
| Ramin Farjadrad | 联合创始人兼 CEO | 130 多项专利发明人;曾联合创立 Velio 和 Aquantia;参与覆盖 Ethernet 与 BoW/OCP 的标准工作 | 互连架构技术叙事和市场叙事的主要负责人 | 很高 |
| Patrick Soheili | 联合创始人兼战略 / 商务负责人 | 曾任 eSilicon、Altera、AMD 和 TRW 高管;具备创投与业务拓展经验 | 商业化、合作伙伴关系和资本市场转译 | 高 |
| Syrus Ziai | 联合创始人兼运营 / 工程负责人 | 曾任 Nuvia、Ikanos、Qualcomm 和 PsiQuantum 工程负责人 | 覆盖封装、架构和多团队工程规模化的执行 | 高 |
| Jason Taylor | 独立董事 | 曾任 Facebook/Meta 基础设施高管,具备 AI 系统和电信背景 | 补充 hyperscale 基础设施视角和标准化可信度 | 中 |
| Umesh Padval | 董事 | Seligman Ventures 管理合伙人;资深半导体投资人兼运营者 | 与 Series C 扩张和融资绑定的后期董事会支持 | 中 |
仅限公开披露的创始人与董事会子集;Eliyan 未公布完整治理文件包或所有权矩阵。
[CO012, CO013, CO014, CO015, CO016, CO017]1.3 融资历史与投资人信号
Eliyan 的资本历史显示,公司从 VC 支持的技术验证快速推进到战略生态融资。2022 年 11 月 Series A 融资 $40M,并与一次 5nm tapeout 搭配披露,把融资和工程里程碑强绑定。2024 年 3 月 Series B 从 Samsung Catalyst Fund 和 Tiger Global 融到 $60M,Intel Capital、SK hynix 等跟投,投资人基础向平台和存储相邻战略方扩展。2026 年 1 月,Eliyan 又宣布来自 AMD、Arm、Coherent、Meta、Samsung Catalyst Fund 和 Intel Capital 的 $50M 战略融资,明确指向商业化和 AI 系统部署。最后,2026 年 7 月 Series C 以 $1B 估值再融 $145M,由 Seligman Ventures 领投,Cisco Investments 和 Lumentum 参投。仅加总公开披露轮次,意味着 2022 年以来公告资本至少 $295M。比算术更重要的是模式:Eliyan 的股东表正在变成 AI 互连栈的地图,横跨计算、代工、存储、网络和光学阵营。[CO009, CO010, CO011, CO020, CO021, CO022]
| 利益相关方 | 角色 | 控制权 / 经济重要性 | 尽调问题 |
|---|---|---|---|
| Seligman Ventures | Series C 领投方 | 领投 $145M Series C,并让 Umesh Padval 进入董事会 | 确认按比例跟投权、治理权和清算优先权 |
| Cisco Investments | Series C 新战略投资方 | 与扩展互连网络相关的网络战略方 | 厘清商业合作与纯财务信号的边界 |
| Lumentum | Series C 新战略投资方 | 光器件战略方,与电光路线图一致 | 验证关系是否包含产品或商业化合作 |
| AMD / Arm / Meta / Coherent | 2026 年 1 月战略投资财团 | 横跨计算和超大规模领域的强生态验证 | 识别其中是否有人是客户、合作伙伴或董事会观察员 |
| Samsung Catalyst Fund 与 Intel Capital | 重复参投的战略投资方 | 多轮回归,意味着持续信心 | 了解排他性、代工或路线图影响 |
| Tiger Global、Tracker Capital 与 Celesta Capital | 早期多轮财务投资方 | 战略生态之外的资本支持 | 评估所有权集中度和退出预期 |
| SK hynix 与 Micron | 内存相关战略投资方 | 释放裸片到内存和 HBM 定位的相关性信号 | 验证互动是否延伸到产品部署或联合设计 |
投资方图谱只反映新闻稿和第三方摘要中公开具名的投资人;权利、持股比例和侧函均未披露。
[CO010, CO020, CO021, CO022, CO023]1.4 里程碑与商业化姿态
公开里程碑显示,Eliyan 已经越过纯概念阶段,但还没有进入充分披露的规模化故事。2022 年,公司宣布 5nm tapeout,并把 Series A 资金与商业就绪挂钩。到 2023 年,NuLink 公开对齐 UCIe 和 BoW,同时 Jason Taylor 加入董事会,释放生态和基础设施相关性增强的信号。2024 年,公司宣布 3nm、每链路 64Gbps 的里程碑、Samsung SF4X 移植,并继续强调相对先进封装的标准封装经济性。2025 年,Eliyan 加入 Arm Total Design 生态,把技术连接到更广泛的开放 chiplet 架构工作。2026 年两次公告最关乎商业化:1 月战略融资称获得 hyperscaler 和 AI 平台投资人的积极支持;7 月 NuLink-XD 发布,则把叙事从封装级 D2D 扩展到面向 scale-up 网络的芯片到模块电光链路。即便如此,公司自己的措辞仍强调客户在评估或部署技术,而不是点名生产收入、出货量或已签 design-win 数量;因此商业化可信,但仍不完全透明。[CO019, CO025, CO026, CO027, CO028, CO029]
| 日期 | 事件 | 类型 | 金额 / 估值 / 状态 | 参与方 | 含义 |
|---|---|---|---|---|---|
| 2022-11-08 | 宣布 Series A 和 5nm 流片 | 融资 | $40M;流片已完成 | Tracker Capital、Celesta、Intel Capital 与 Micron | 融资与首个公开商业就绪里程碑绑定 |
| 2023-08-24 | 宣布支持 UCIe 兼容 | 产品 | UCIe / BoW 对齐 | Eliyan、UCIe 生态 | 把公司放进开放 chiplet 标准路径 |
| 2023-11-28 | Jason Taylor 加入董事会 | 治理 | 董事会扩容 | Eliyan、Jason Taylor | 补充超大规模基础设施经验 |
| 2024-03-25 | 宣布 Series B | 融资 | $60M | Samsung Catalyst Fund、Tiger Global、Intel Capital、SK hynix 等 | 扩大战略和财务投资人基础 |
| 2024-10-09 | 宣布 64Gbps 3nm chiplet PHY 里程碑 | 产品 | 基于标准封装的 NuLink-2.0 演示载体 | Eliyan | 显示高性能 D2D 路线图继续推进 |
| 2024-11-20 | 宣布 Samsung SF4X 移植和 HBM4 base-die 意向 | 产品 | SF4X 上最高 40Gbps | Eliyan、Samsung Foundry | 将路线图延伸到近内存和特定代工平台用例 |
| 2025-11-11 | 宣布 Arm Total Design 生态合作 | 合作 | 支持 FCSA | Eliyan、Arm | 提升其在开放 chiplet 架构中的生态定位 |
| 2026-01-28 | 宣布 $50M 战略融资 | 融资 | $50M | AMD、Arm、Coherent、Meta、Samsung Catalyst Fund 与 Intel Capital | 释放来自超大规模云厂商和计算生态的商业化支持信号 |
| 2026-07-24 | 推出 NuLink-XD 224G PAM4 SerDes | 产品 | 聚焦光模块和扩展互连 | Eliyan | 从封装互连扩展到电光扩展链路 |
| 2026-07-29 | 以独角兽估值完成 Series C | 融资 | $145M,估值 $1B | Seligman Ventures、Cisco Investments 与 Lumentum | 将 Eliyan 定位为资本更充足的 AI 互连平台 |
基于已审阅官方发布和相关新闻报道建立的单一公开时间线;时间线不包含未披露的内部里程碑和客户特定事件。
[CO009, CO010, CO019, CO020, CO021, CO022]Eliyan 的公开叙事从流片和标准对齐,推进到生态融资与电光互连扩展。
[CO009, CO020, CO021, CO022, CO026, CO027]1.5 风险、依赖集中与披露缺口
公司概览层面的核心风险,不在于 Eliyan 是否有一套技术上有意思的互连架构,而在于这套架构能否足够快地进入广泛量产,在更大生态玩家把类似能力内化之前支撑独角兽估值逻辑。Eliyan 自有材料强调标准封装的成本和供应链优势,但负面行业文献仍指向 chiplet 系统中未解决的封装、测试、互操作、安全、热管理和商业化挑战。EE Times 认为,市场意识到并非每个点状方案都能作为独立公司存活后,该行业正进入整合阶段;SemiEngineering 则强调,AI 基础设施站在爆发需求和成本、复杂度、部署约束堆叠的十字路口。Eliyan 仍是一家披露轻的私营公司。已审阅的公开材料没有给出经审计收入、ARR、客户集中度、完整董事会机制或权威所有权数据;即便员工数也只能通过 Tracxn 等第三方估算获得。这意味着投资人能看到动能和生态验证,但还无法用标准经营指标披露来三角校验估值。[CO032, CO034, CO035, CO036, CO037, CO038]
1.6 展示材料
02市场分析
2.1 市场边界与现状替代方案
Eliyan 不参与整个半导体市场,甚至也不参与整个 AI 基础设施市场。它的直接收入池位于让多裸片 AI 系统彼此通信的层:裸片到裸片 PHY IP、存储互连逻辑、芯片到芯片链路、相关 chiplet,以及相邻的电光连接。这比广义 chiplet 市场更窄,后者包括许多计算和子系统 chiplet;也比先进封装支出更窄,后者包括 Eliyan 不直接变现的代工、OSAT 和基板价值。因此,正确的市场边界应从 chiplet 互连和 UCIe 式连接开始,再扩展到相邻替代预算,例如先进封装、专有裸片到裸片链路、机架级 fabric 交换机、交换侧共封装光学,以及光子互连引擎。这个区分重要,因为标题式 TAM 可以很大,但仍会夸大 Eliyan 近期可现实变现的细分市场。实践中,相关买家问的是要不要把预算投向 AI 加速器和存储系统周边更好的连接,而不是抽象地要不要投向半导体。[CM001, CM002, CM003, CM004, CM015, CM037]
| 细分市场 / 类别 | 纳入支出 | 排除支出 | 买方 / 付款方 | 与 Eliyan 的相关性 |
|---|---|---|---|---|
| Chiplet 互连 / UCIe 生态 | 裸片间 PHY IP、控制器、互操作层、验证及相关生态服务 | 大多数加速器硅价值、服务器 BOM 和通用网络支出 | ASIC 团队、封装架构师、超大规模云厂商、平台设计方 | 观察 Eliyan 核心产品最贴近的公开市场镜头 |
| 更广义 chiplet 市场 | 所有支持 chiplet 的硅构建模块和模块化 IC 内容 | 通用 IT 基础设施和非 chiplet 半导体 | 硅供应商和系统架构师 | 有助于建立背景,但宽于 Eliyan 直接变现范围 |
| 先进封装市场 | 中介层、桥接、基板、组装、测试和封装服务 | 大多数纯 IP 和软件价值 | 晶圆代工厂、OSAT、先进封装买方 | 重要相邻市场和约束层,不是 Eliyan 直接收入池 |
| 机架级 AI 连接 | 互连网络交换机、重定时器、信号调理器、光引擎、电路交换、近封装光学 | 与连接无关的通用计算硅 | 机架架构师、网络平台团队、超大规模云厂商 | 直接替代品和相邻预算类别 |
| 现状替代方案 | 专有 D2D 链路、可插拔光模块、交换机侧 CPO、传统重 SerDes 设计 | 供应商中立 chiplet IP 和内存扩展 chiplet | 既有硅 / 平台所有者 | 宽口径 TAM 不会一比一转成 Eliyan 收入的主要原因 |
各行把 Eliyan 贴近产品的连接层,与更广义的 chiplet、封装以及相邻光学 / 网络支出区分开。
[CM001, CM002, CM003, CM004, CM015, CM037]Eliyan 的直接机会嵌在更大的 chiplet 和先进封装支出池里。
[CM006, CM008, CM009, CM010, CM015]2.2 规模测算视角——TAM、SAM 与贴近产品的市场代理
公开市场规模估算分歧很大,因为它们测量的边界不同。Fortune Business Insights 将 2026 年 chiplet 互连和 UCIe 生态规模估为 $3.28B,并预测多年高速增长;GM Insights 将 2026 年 chiplet 互连市场放在 $2.89B;Intel Market Research 的更广义 chiplet 市场口径在 2026 年升至 $7.31B。PatSnap 又用更宽口径,把 2026 年先进封装市场放在约 $49B–$55B。这些数字并不互相矛盾,更像嵌套口径。Eliyan 更接近互连和标准层,而不是整个 chiplet 栈;距离完整先进封装经济性更远。公司同时跨越电连接和光学相邻连接,因此比纯 UCIe-IP 厂商更宽,但比完整机架级网络公司更窄。对估值最有用的结论不是一个精确 TAM,而是一组市场区间:今天直接互连机会为低个位数十亿美元,更广 chiplet 背景为高个位数十亿美元,封装与光学相邻空间则大得多,但只有一部分会转化为 Eliyan 可寻址收入。[CM005, CM006, CM007, CM008, CM009, CM010]
| 发布方 | 年份 | 地域 | 市场定义 | 数值 / 区间 | 增长信号 | 方法说明 | 置信度 | 关键限制 |
|---|---|---|---|---|---|---|---|---|
| Fortune Business Insights | 2026 | 全球 | Chiplet 互连和 UCIe 生态 | 2026 年 $3.28B;2034 年 $23.47B | 到 2034 年 CAGR 27.9% | 专门的 UCIe 生态视角 | 中 | 商业研究机构;宽于 Eliyan 直接产品范围 |
| GM Insights | 2026 | 全球 | Chiplet 互连市场 | 2026 年 $2.89B;2031 年 $12.42B | 到 2035 年 CAGR 34.4% | 聚焦互连的品类市场视角 | 中 | 边界不同于 Fortune,可能纳入不同层 |
| Intel Market Research(市场研究) | 2026 | 全球 | 更广义 chiplet 市场 | 2026 年 $7.31B;2034 年 $18.15B | 到 2034 年 CAGR 13.2% | 更广义模块化半导体视角 | 低-中 | 对 Eliyan 直接 SAM 来说过宽 |
| PatSnap | 2026 | 全球 | 先进封装市场 | 到 2026 年 $49B-$55B | 结构性拐点 | 覆盖封装生态的相邻视角 | 低-中 | 不是 Eliyan 直接收入市场 |
| Fortune Business Insights | 2026 | 全球 | 互连栈内的 UCIe 增长 | 增长最快的子板块 | UCIe CAGR 44.2% | 开放标准子板块估计 | 中 | 子板块增长不等于 Eliyan 份额 |
| 作者综合 | 2026 | 全球 | 贴近 Eliyan 产品的市场边界 | 低个位数十亿美元 | 嵌在更广义 chiplet 和封装总量之内 | 仅围绕连接层综合公开视角 | 低 | 不是正式分析师估计,也不是营收预测 |
规模视角有意做成嵌套:chiplet 互连是最接近的公开代理,更广义 chiplet 提供背景,先进封装是相邻市场。
[CM005, CM006, CM007, CM008, CM009, CM010]公开的 2026 年市场口径显示,互连集中在低个位数十亿美元,广义 chiplet 和封装更高。
[CM006, CM008, CM009, CM010]2.3 买家、用户与付费方结构
Eliyan 这类连接产品的直接买家很少是独立采购团队。更常见的是一条角色链:从 AI 加速器架构师、封装设计团队、存储接口团队、板级或机架设计者,到设定系统级带宽和功耗目标的 hyperscaler 基础设施团队。Eliyan 自有产品材料和融资公告指向 AI 加速器、存储扩展、云基础设施和高性能计算,说明这些就是即时问题集。周边市场也证实,预算越来越由做系统架构决策的团队掌握。Astera Labs 把机架定义为 AI 基础设施的新计算单元,Broadcom 围绕能效和带宽密度定义共封装光学,Lightmatter 和 Ayar Labs 都在销售旨在解决类似集群级瓶颈的光互连系统。这意味着买方控制权在平台架构师和战略硅团队手中,而不是在商品化模块采购手中。用户和付费方也可能不同:hyperscaler 可能定义工作负载需求,但最终由 ASIC 团队、交换机厂商或封装伙伴决定哪套互连栈被设计进去。[CM016, CM017, CM018, CM019, CM020, CM021]
| 细分市场 | 买方 | 用户 | 付款方 | 工作流 / 预算负责人 | 采用触发 |
|---|---|---|---|---|---|
| AI 加速器 D2D / D2M | ASIC 设计团队 | 系统架构师和内存团队 | 超大规模云厂商或 OEM 平台预算 | 硅架构主导权 | 每个封装需要更高带宽或更多内存 |
| 机架级 AI 横向扩展互连 | 机架 / 网络架构团队 | 集群运营方 | 超大规模云厂商基础设施资本开支 | 基础设施与网络架构 | 需要突破服务器级带宽上限 |
| 内存扩展 / HBM 邻近设计 | 内存接口与封装团队 | 加速器项目负责人 | 硅与平台预算 | 封装、内存与硅协同设计团队 | 不靠单片裸片变大也要增加内存容量 |
| 光互连邻近的芯片到模块链路 | 光模块与系统团队 | 网络 / 系统集成团队 | 网络或平台资本开支 | 互连平台负责人 | 需要比电链路更低功耗、更长距离 |
| HPC / 云平台设计 | 计算平台团队 | 基础设施工程 | 数据中心平台预算 | 跨职能架构决策 | 需要从 chiplet 拿到延迟、功耗或模块化收益 |
买方、用户和付款方常常不是同一方,因为互连选择发生在架构项目里,而不是商品化采购流程里。
[CM016, CM017, CM018, CM019, CM020, CM021]Eliyan 这类互连决策由架构链条决定,而不是单一买方拍板。
[CM016, CM017, CM018, CM036, CM040]2.4 增长驱动因素
最主要的市场驱动很简单:AI 系统碰到存储和 I/O 上限的速度,快过原始算力提升能够弥补的速度。Eliyan 自有材料把这些称为存储墙和 I/O 墙,Electronic Design 与 SemiAnalysis 也强化了同一个结构性判断:AI 工作负载需要更多带宽、更低延迟、更大存储容量和更好能效。开放标准是第二个驱动。Fortune 和 UCIe Consortium 都显示,生态正在把 UCIe 推向默认互操作层,降低多供应商 chiplet 架构的摩擦。第三个驱动是优化重点从单节点迁移到机架级或集群级。Astera Labs、Broadcom、Lightmatter、Ayar Labs、Celestial AI,以及 Marvell-Lumentum 光网络演示,都显示大厂正在围绕同一瓶颈投资。即便 Eliyan 不能赢下每个预算类别,也会受益于这种结构性需求。最后,投资人行为本身也是市场信号:Eliyan 来自计算、hyperscaler、存储、网络和光学参与者的战略轮融资表明,主要生态玩家认为互连正从后台实现细节变成定义性能的层。[CM025, CM026, CM027, CM028, CM029, CM030]
| 驱动因素 / 约束 | 方向 | 时间 | 影响 | 尽调追问 |
|---|---|---|---|---|
| 内存墙与 I/O 墙 | 正向 | 当前 | 加速器和 HBM 周边连接需要升级,形成结构性需求 | 量化哪些工作负载最需要 Eliyan,而不是替代方案 |
| UCIe 与开放标准势能 | 正向 | 当前至多年期 | 多供应商 chiplet 设计的生态摩擦降低 | 确认买方要的是标准 UCIe,还是专有增强 |
| 机架级 AI 架构 | 正向 | 当前至多年期 | 预算转向系统级连接选择 | 梳理 Eliyan 相比机架交换机和 retimer 厂商的位置 |
| 战略生态投资 | 正向 | 当前 | 说明市场相信互连能创造增量价值 | 核实投资方是否也是设计合作伙伴或客户 |
| 封装与布线复杂度 | 负向 | 当前 | 可能拖慢设计周期,压缩部署窗口 | 评估晶圆厂 / OSAT 约束与封装依赖 |
| 测试 / 互操作 / 安全挑战 | 负向 | 当前至多年期 | 可能推迟量产,并利好更大的既有厂商 | 索取认证与已知良品裸片数据 |
| 来自光互连和机架既有厂商的替代压力 | 负向 | 当前 | 解决同一瓶颈的邻近厂商可能用不同路径拿走预算 | 明确 Eliyan 相比 Astera、Broadcom、Lightmatter、Ayar 以及 Marvell 系光互连栈的定位 |
| 缺少公开定价和赢单数量披露 | 负向 | 当前 | 让 SOM 和采用曲线建模不够精确 | 索取定价架构、管线阶段,以及每个设计赢单的预期内容价值 |
同一组宏观力量既创造 TAM,也抬高集成负担;本表把结构性需求和拖慢转化的门槛摩擦拆开。
[CM025, CM026, CM027, CM028, CM029, CM030]采用从生态需求推进到架构认证,再到量产部署。
[CM025, CM029, CM033, CM035, CM038]2.5 约束、采用摩擦与争议
有吸引力的需求背景并不会消除采用摩擦。SemiEngineering 对 chiplet 的报道强调架构复杂度上升,其存储墙报道也说明,即便需求明确,封装、布线和存储集成仍然很难。EE Times 给出更明确的负面市场观点,认为 chiplet 格局正走向整合,测试、互操作、安全和光学转型等未解决问题将决定哪些厂商能作为独立公司活下来。这些约束直接作用于 Eliyan,因为它的最佳叙事一次横跨多个领域:裸片到裸片、裸片到存储、芯片到芯片,以及电光 scale-up。每增加一个领域都会扩大 TAM,但也提高集成范围和买方认证负担。替代风险也存在。大型相邻厂商已经在销售或营销机架级 AI 交换机、光引擎、CPO 或电路交换 fabric,它们能从不同角度抓住同一条预算线。结果是:市场需求强、紧迫性真实,但认证周期仍长,技术相关性和收入兑现之间仍有明显距离。[CM033, CM034, CM035, CM036, CM037, CM038]
2.6 展示材料
03竞争对手
3.1 格局与直接同业
Eliyan 不在广义 AI 芯片市场竞争;它竞争的是更窄但战略重要的层,负责在裸片、存储子系统和日益增多的光链路之间搬运数据。公开证据显示,至少有四类对手重要。第一类是 Astera Labs、Alphawave 等直接 D2D 和 UCIe 导向供应商,它们用 IP 或硅产品解决先进封装和裸片到裸片连接。第二类是 Broadcom、Marvell 等大型在位者,它们可以把互连技术挂到更广的网络、光学或定制硅业务上。第三类是 Lightmatter、Celestial AI、Ayar Labs 等光子专业厂商;当连接从铜进一步外移时,它们的竞争更直接。最后,对最高出货量买家来说,hyperscaler 内部自研仍是替代方案。这意味着 Eliyan 的竞争集合不是静态的:有些对手攻击封装级链路,有些攻击机架级光学,有些两边都能做。实际含义是,Eliyan 必须证明其特定接口方案在哪些地方不可替代,而不只是符合标准。[CP001, CP002, CP003, CP004, CP005, CP006]
| 竞争者 | 类别 | 当前切入点 | 有证据支撑的优势 | 局限 / 缺口 |
|---|---|---|---|---|
| Eliyan | 直接专业厂商 | PHY IP、chiplet、内存和光互连路线图 | 聚焦 chiplet 带宽密度和内存墙用例 | 公开客户证据仍薄 |
| Astera Labs | 直接 / 邻近 | 面向 AI 基础设施的连接硅和系统组件 | 平台宽度和上市公司规模已经验证 | 并非只聚焦 Eliyan 所在的 D2D 细分位点 |
| Alphawave Semi | 直接 IP 同行 | 先进节点 UCIe IP 与 chiplet 连接 | 已展示 3nm 上 Gen3 UCIe IP 进展 | 已检索来源中的公开采用细节有限 |
| Broadcom | 邻近既有厂商 | 网络、光互连和共封装光学栈 | 已装机基础和广泛打包能力 | 对 Eliyan 式内存链路的公开表述不够具体 |
| Marvell | 邻近既有厂商 | 定制硅、光互连,以及与 NVIDIA 生态对齐 | 覆盖 AI 基础设施多类平台 | 以更大的系统级厂商身份竞争,而不只是 D2D |
| Lightmatter / Celestial / Ayar | 光互连邻近厂商 | 光子连接和内存 / 网络互连 | 若价值继续转向封装外,契合度高 | 已检索材料中的公开部署细节不完整 |
各行比较的是买方流程中的角色,而不是在证据不足时强排收入份额。
[CP002, CP003, CP004, CP005, CP006, CP007]Eliyan 最接近 chiplet 专精和新兴光学相邻市场的交点,既有厂商则偏向平台宽度。
[CP001, CP002, CP003, CP007, CP008, CP019]3.2 能力与买方比较
最有用的比较不是公司规模,而是每家厂商处在买方工作流的哪个位置。Eliyan 的公开材料强调 PHY IP、chiplet、存储链路和具体带宽密度指标;几家竞争对手则从相邻切口切入同一痛点:经过验证的板级连接平台、可互操作的 UCIe IP 模块,或面向更远距离数据移动的光子 fabric。因此,买家选择的是架构,不只是品牌。在封装内或存储墙附近优化的系统团队,可能更偏好能调校 D2D 和面向存储链路的供应商。设计更广 AI fabric 的团队,则可能更看重生态集成和光学路线图。公开证据在能力上远强于在标价上,因此比较必须围绕集成负担、覆盖距离、功耗和生态契合来展开。这也意味着,没有支持证据的单元格应保持未知,而不是从营销材料里猜。[CP009, CP010, CP012, CP014, CP015, CP021]
| 购买标准 | Eliyan | Astera / Alphawave | Broadcom / Marvell | 光互连邻近厂商 |
|---|---|---|---|---|
| 封装级裸片到裸片聚焦度 | 强 | 强 | 中 | 中 |
| 内存墙定位 | 强 | 中 | 中 | 中 |
| 机架级光互连邻近性 | 起步 | 低至中 | 强 | 强 |
| 广泛生态打包 | 中 | 中 | 强 | 中 |
| 公开互操作信号 | 中 | 中至强 | 强 | 中 |
序数评分概括已检索材料中的明确定位;证据不足的单元格保持粗粒度。
[CP009, CP010, CP012, CP014, CP028, CP029]| 厂商类型 | 公开定价可见度 | 已观察到的合同模式 | 可能的切换成本驱动 | 影响 |
|---|---|---|---|---|
| Eliyan | 低 | 定制设计导入 / 半导体项目合作 | 集成与认证投入 | 难以按标价对标 |
| Astera / 平台硅 | 低 | 平台或组件附加销售 | 已装机基础验证 | 可能受益于打包经济性 |
| IP 供应商 | 低 | 授权加实施支持 | 设计流程兼容性 | 比较取决于工程场景 |
| 光互连邻近厂商 | 低 | 按项目定制 | 系统架构承诺 | 采用取决于更大的拓扑选择 |
公开来源没有披露各同行的标价。
[CP015, CP021, CP035]公开证据更支持按能力领域对比,而不是按价格对比。
[CP010, CP012, CP014, CP028, CP032]3.3 分销、切换成本与替代方案
互连基础设施的竞争结果不只取决于技术优劣,因为认证、封装和生态对齐会制造很长的设计周期。上市公司在位者可以把连接与交换机、NIC、光学或定制硅项目打包,从而在采购中获得杠杆,即便它们没有证明某一个链路指标绝对最优。Eliyan 来自 hyperscaler 和基础设施伙伴的战略投资有助于抵消这种失衡,但不能抹掉它。现状替代方案尤其重要:最大客户可以把互连设计内部化,或从平台厂商采购半定制打包方案。切换成本看起来更多来自设计返工、验证工作和生产风险容忍度,而不是公开标价。因此,design-win 时间和生态伙伴访问能力,是竞争战的核心部分。Eliyan 的机会在于定制 AI 架构越来越需要差异化连接;挑战在于,同一趋势也让内部自研和在位者打包更可信。[CP016, CP017, CP018, CP020, CP023, CP024]
| 因素 | 已观察证据 | 受益方 | 对 Eliyan 的风险 | 尽调追问 |
|---|---|---|---|---|
| 战略投资方 | 超大规模云厂商和基础设施战略资金 | Eliyan | 有帮助,但不是确定性渠道证据 | 追问实际设计赢单贡献 |
| 既有厂商打包能力 | Broadcom/Marvell/NVIDIA 生态宽度 | 既有厂商 | 价格与采购压力 | 追问替换案例 |
| 内部自研选项 | 定制 AI 基础设施项目 | 大客户 | 可能压低创业公司在客户预算中的份额 | 追问自研还是外采的决策逻辑 |
| 认证负担 | Chiplet / 封装复杂度 | 已装机厂商 | 即使规格更好,也可能拖慢切换 | 追问集成周期 |
| 生态伙伴关系 | Arm 和标准参与 | 混合 | 必要但不足以形成差异化 | 追问伙伴带来的管线证据 |
半导体分销力通常不是直接销售,而是通过设计项目和生态契合间接体现。
[CP016, CP017, CP018, CP020, CP023, CP031]Eliyan 的公开优势在技术具体性上最强,在持久锁定的公开证据上最弱。
[CP017, CP024, CP025, CP037, CP038]3.4 护城河持久性与替代风险
当前公开记录更能支持 Eliyan 的技术雄心,而不是持久商业锁定。护城河似乎建立在技术定位、标准执行、伙伴访问和及时切入存储及电光瓶颈上。这有意义,但不同于已证明的装机基础或客户留存引擎。风险是双向的。一边,更广泛的标准采用和资金更充足的 UCIe 对手,可能把今天看起来独特的接口主张商品化。另一边,随着光学和机架级 fabric 成熟,一部分价值可能迁移到封装之外,降低纯电互连供应商可捕获份额。EE Times 的整合视角在这里重要:战略意义足够大的类别,往往会收敛到少数规模化赢家,尤其是在客户偏好经过验证的多产品供应商时。因此,Eliyan 看起来有差异化,但尚未被保护起来。承销问题是,它能否在在位者和光学打包方案缩小差距前,把技术特殊性转化为重复 design-win。[CP019, CP022, CP025, CP026, CP030, CP034]
3.5 展示材料
04财务
4.1 收入模式与披露限制
Eliyan 的公开材料足以清楚描述它卖什么,但不足以量化它赚多少、利润率如何。公司看起来通过互连 IP、PHY 技术、chiplet,以及面向 AI 和高性能计算系统的相关 design-in 项目来变现连接技术。这足以理解商业模式类别:Eliyan 把基础设施组件卖进长周期半导体项目,而不是面向消费者的软件或商品化硬件。仍不透明的是,授权、硅含量、支持和定制工程之间的收入结构。已检索到的 2026 年融资或媒体材料,没有任何一项公布收入、毛利率、ARR、出货量或客户集中度。因此,仅靠公开证据无法判断收入质量。公司的信息披露优先强调技术性能、生态定位和未来商业化,这符合一家仍在证明从设计相关性走向规模化变现路径的 VC 支持半导体基础设施公司。[CI001, CI002, CI003, CI007, CI008, CI023]
| 收入流 | 机制 | 公开状态 | 收入质量判断 | 尽调追问 |
|---|---|---|---|---|
| PHY / 互连 IP | 授权或设计导入经济性 | 有定性支撑 | 若可复用,可能有吸引力,但尚未量化 | 索取授权模式和版税结构 |
| Chiplet / 连接硅 | 客户项目中的硅内容价值 | 有定性支撑 | 可能随出货系统增长,但单元数未知 | 索取已出货项目数量和 ASP |
| 定制工程 / 集成 | 按项目提供支持 | 仅为推断 | 可能带来不均衡的里程碑收入 | 索取服务收入占比和确认政策 |
| 光互连路线图产品 | 未来产品扩展 | 仅路线图 | 现在纳入模型还太早 | 索取商业化发布时间和附加率 |
公开证据支持商业模式类别,但不支持当前构成或实际变现水平。
[CI001, CI002, CI007, CI034, CI035]| 问题 | 公开答案 | 置信度 | 重要性 | 来源 / 缺口 |
|---|---|---|---|---|
| 能看到标价吗? | 否 | 高 | 没有价格点,预测只能从设计赢单出发,而不是产品目录经济性 | 融资和产品页面没有定价 |
| 能看到版税率吗? | 否 | 高 | IP 收入质量取决于费率结构 | 未检索到公开合同案例 |
| 能看到硅 ASP 吗? | 否 | 高 | 硬件毛利建模需要该数据 | 未披露出货量或 ASP |
| 能看到折扣 / 打包结构吗? | 否 | 高 | 与既有厂商竞争时很关键 | 没有公开客户合同 |
所有变现单元格仍是公开数据缺口,不能作为模型输入。
[CI003, CI008, CI031, CI032]Eliyan 似乎把客户架构痛点转化为 IP、硅片和集成经济性的组合,但实际定价仍未披露。
这张桥图是结构性而非数字化的,因为公开材料未披露已确认收入、版税或 ASP。
[CI001, CI002, CI007, CI008]4.2 销售动作与成本结构
Eliyan 的财务机制很可能与软件公司有本质差异,因为这个类别靠设计周期、验证工作和生态认证来销售。公开证据显示,销售效率将由少数大型技术项目塑造,而不是由高速获客驱动。战略投资人可能带来渠道,但不能消除采购风险,也不能证明需求可复现。成本侧,已检索来源支持一个方向性但非数字化结论:chiplet 和互连商业化资本强度高。代工、封装、集成、验证和现场支持,负担都可能比纯 IP 故事暗示的更重。Eliyan 的光学扩展会在相关收入变得可见之前进一步放大负担。招聘页活动也说明固定运营成本仍在累积。不过,没有公开来源提供 CAC、回本周期、毛利率或贡献利润率,因此真正的经济引擎仍藏在技术叙事之后。[CI009, CI010, CI014, CI015, CI016, CI022]
| 指标 | 公开值 | 置信度 | 重要性 | 尽调追问 |
|---|---|---|---|---|
| CAC / 回本周期 | null | 低 | 长设计周期可能掩盖销售效率偏弱 | 追问漏斗、周期,以及分细分市场的转化 |
| 毛利率 | null | 低 | 对拆分硬件与 IP 组合至关重要 | 要求按产品系列披露利润率 |
| 贡献利润率 | null | 低 | 测算经营杠杆所必需 | 要求说明支持与服务成本负担 |
| 营运资本强度 | null | 低 | 硬件放量可能在确认收入前先吃掉现金 | 要求披露库存、采购承诺和应收账款条款 |
| 单项目收入 | null | 低 | 估值需要映射到销售管线厚度 | 要求披露积压订单和已锁定设计定点的经济性 |
这些空值是有意保留;公开资料未披露单位经济指标字段。
[CI009, CI015, CI016, CI017, CI022]公开证据展示了成本栈逻辑,但缺少收口这座桥所需的公司专属指标。
公开同行文件有可比披露,但 Eliyan 未发布自身利润率、现金消耗或 CAC 数据。
[CI009, CI014, CI015, CI016, CI022]公开证据意味着商业化需求很重,但实际现金转化披露很少。
[CI004, CI016, CI022, CI033, CI036]4.3 资本充足性与融资依赖
公开可见的是融资栈。Eliyan 在 2026 年 1 月宣布 $50M 战略融资,2026 年 7 月宣布 $145M Series C,叠加此前 Series A 和 Series B。披露时间线意味着公告资本至少 $295M,对一家年轻半导体公司而言很有分量。新资金被一致描述为商业化和 scale-up 资本,而不是紧急资本重组,这是正面信号。与此同时,融资节奏显示公司仍处于积极建设期,因此在收入规模化之前依赖外部资本。公开记录没有披露账上现金、月度 burn、runway、债务或采购义务,因此无法精确建模偿付能力。最稳妥的结论是,2026 年融资后近期融资风险看起来降低,但估值压力提高,因为轮次规模和独角兽标记要求技术承诺快速转化为市场牵引。[CI004, CI005, CI006, CI011, CI012, CI018]
| 项目 | 公开证据 | 状态 | 含义 | 尽调追问 |
|---|---|---|---|---|
| Series A 轮 | 宣布融资 40M | 历史事项 | 早期产品建设资金可见 | None |
| Series B 轮 | 宣布融资 60M | 历史事项 | 支撑 2026 年发力前的扩张 | None |
| 战略融资 | 2026 年 1 月宣布融资 50M | 当前 / 近期 | 显示生态站队与商业化投入 | 追问投资方是否也是客户 |
| Series C 轮 | 2026 年 7 月以 $1B 估值宣布融资 145M | 当前 / 近期 | 缓解近期融资压力,但也抬高业绩兑现预期 | 追问投后股权和清算条款 |
| 账面现金 | 未披露 | 缺口 | 无法直接测算资金续航期 | 要求提供最新现金余额 |
| 债务 / 项目融资 | 未披露 | 缺口 | 资产负债表风险无法排除 | 要求提供所有授信 / 融资安排及限制性条款 |
本表关注前瞻性资本充足性,而不是重列所有历史融资细节。
[CI004, CI005, CI006, CI011, CI012, CI018]最清晰的公开数字输入是融资轮次,而不是经营指标。
该图使用已公告的轮次规模,因为收入和资金续航未披露。
[CI011, CI012, CI029]4.4 财务结论与尽调阻塞项
公开证据支持一个狭义财务结论:相对于 Eliyan 选择披露的信息,它看起来融资充足;但若要完整承销,披露仍严重不足。正面项很清楚。独立报道和公司公告持续记录强劲投资人胃口、庞大的 2026 年融资栈,以及一条如果 design-win 能转化为可重复收入、就可能支撑长期价值的路线图。负面项同样清楚。没有公开收入基础、没有毛利率披露、没有 burn 或 runway 数字、没有营运资本可见度,也没有客户集中度或续约动态证据。上市公司可比对象展示了成熟基础设施厂商通常会有怎样的披露,也反衬这里缺了多少。对投资人来说,关键尽调问题不是 Eliyan 有没有融到钱,而是这些钱是否足够快地转化为商业上持久的项目,足以支撑 $1B 估值且无需再进行高度稀释性的融资。[CI013, CI020, CI021, CI024, CI025, CI036]
4.5 展示材料
05产品与技术
5.1 产品定义与模块地图
Eliyan 的产品表面最好理解为使能型基础设施,而不是独立计算产品。公司销售的是连接组织,帮助先进 AI 和 HPC 系统在裸片、存储子系统以及越来越多的光相邻链路之间搬运数据。公开材料始终指向一组构件——互连 IP、PHY 技术、chiplet,以及 UMI 等存储扩展结构——而不是单一整体 SKU。这个区分重要,因为真正买家不是通用 IT 采购,而是 OEM、hyperscaler 或硅项目内部的系统架构 / 封装团队。Eliyan 的模块叙事足以做概念映射,但还没有形成公开数据表非常丰富的完整表面。投资人可以验证架构方向,但仍缺少完整公开封装清单、支持矩阵,或每个声称组件的部署证据。因此,本章把产品定义主张视为支撑充分,把模块完整性主张视为部分支撑。[CE001, CE002, CE003, CE011, CE012, CE023]
| 模块 / 资产 | 主要用户 | 状态 / 成熟度 | 差异化 | 尽调缺口 |
|---|---|---|---|---|
| 互连 IP / PHY | 芯片架构团队 | 公开资料有支撑 | 核心连接模块 | 需要合同和部署细节 |
| NuGear / chiplet 系列 | 系统和封装团队 | 暗示已进入商业化阶段 | 从 IP 走向产品化构件 | 需要量产设计定点证据 |
| UMI / 内存链路概念 | 内存和平台团队 | 技术材料有支撑 | 直接瞄准内存墙扩展 | 需要更广互操作细节 |
| 光互连相邻路线图 | AI 基础设施和模块团队 | 新兴路线图 | 把适用范围从封装内链路往外延伸 | 需要外部依赖图谱 |
各行概括公开可见的产品界面,不应视为完整 SKU 目录。
[CE001, CE003, CE007, CE012, CE024]| 用户任务 | 当前瓶颈 | Eliyan 方案 | 可衡量收益 | 局限 |
|---|---|---|---|---|
| 扩展裸片间带宽 | 封装边缘空间和带宽密度 | 互连 IP / PHY | 单位封装边缘吞吐可能更高 | 未公开跨客户基准测试 |
| 扩展内存带宽 | 内存墙约束 | UMI / 面向内存的链路 | 内存扩展性更好 | 量产证据有限 |
| 为光互连相邻扩展做准备 | 铜互连距离和系统级约束 | NuLink-XD 路线图 | 未来芯片到模块路径 | 仍偏路线图 |
| 在开放生态中集成 chiplet | 互操作摩擦 | UCIe 支持叠加 Arm 生态适配 | 降低生态摩擦 | 仍取决于客户集成 |
用例按客户工作流映射,而不是按营销口号罗列。
[CE002, CE011, CE013, CE014, CE017]Eliyan 的栈从信号和 PHY 层一路延伸到封装、内存,以及光学相邻部署。
[CE001, CE003, CE004, CE016]客户不是把产品当商品化部件采购,而是在架构和封装决策中嵌入它。
[CE002, CE011, CE013, CE017]5.2 架构与差异化
Eliyan 技术故事的核心,是一套围绕存储和带宽的架构判断。它不是让客户只靠更大的单片裸片或通用板级连接来解决扩展问题,而是聚焦在裸片到裸片链路、封装和存储系统变成瓶颈的交汇处。对白皮书披露而言,这家私营半导体创业公司的公开表面异常有帮助:公司发布过封装和同步双向信号相关工作,也公开围绕 UMI 和高性能 PHY 发声。这不能证明市场成功,但意味着外部能以比典型营销网站更深的层级检查其架构。关于存储墙和封装复杂度的独立评论也强化了问题真实存在。差异化因此不只是协议支持,而是 Eliyan 声称能在 chiplet 系统中交付更高带宽密度和更好的面向存储行为。[CE004, CE005, CE014, CE021, CE026, CE029]
| 层 / 组件 | 作用 | 依赖 | 风险 |
|---|---|---|---|
| PHY / 信令层 | 在裸片或模块之间传输数据 | 晶圆厂实现与验证 | 性能宣称可能跑在采用之前 |
| 封装交互 | 决定实际密度和集成难度 | OSAT / 封装设计复杂度 | 进度和良率风险 |
| 内存接口 / UMI | 解决内存墙瓶颈 | 标准和客户架构适配 | 互操作和举证压力 |
| 标准层 | 提升生态适配 | UCIe 演进与合规 | 同质化风险 |
| 光互连相邻延展 | 扩大未来触达范围 | 模块和生态成熟度 | 依赖扩散 |
此处把架构表示为依赖栈,而非完整原理图。
[CE004, CE005, CE008, CE014, CE016, CE026]路线图能否跑通,取决于标准、晶圆厂、封装以及相邻生态参与方。
[CE008, CE009, CE016, CE017, CE028, CE029]公开证据在核心架构上最强,在量产支持证明上最弱。
[CE018, CE019, CE022, CE025, CE031, CE035]5.3 部署、路线图与依赖
公开证据显示,Eliyan 正从技术证明走向更广商业化,但依赖图谱很重。3nm 中实现 64Gbps 的公告、Samsung 可移植工作、最新 UCIe 支持和加入 Arm 生态,都通过展示多层执行来增强路线图可信度:性能、工艺可移植性、标准兼容性和伙伴对齐。随后 NuLink-XD 把路线图延伸到芯片到模块和光相邻机会,可能显著扩大公司的战略相关性。代价是,每一步都增加对代工厂、封装、标准组织和相邻生态玩家的依赖。光学扩展尤其会扩大 Eliyan 无法完全控制的外部就绪因素集合。面向商业化的融资支持公司正准备部署这一判断,但公开记录仍未证明其在具名客户项目中具备广泛量产成熟度。[CE006, CE007, CE008, CE009, CE015, CE016]
| 控制 / 证明 | 公开状态 | 范围 | 缺口 |
|---|---|---|---|
| UCIe 标准支持 | 可见 | 互操作信号 | 不等于量产资格认证 |
| Arm 生态参与 | 可见 | 平台集成信号 | 不能证明现场可靠性 |
| 安全或质量认证 | 未见 | Unknown | 需要 ISO / 安全 / 客户认证数据 |
| 支持 SLA / 可靠性指标 | 未见 | Unknown | 需要缺陷、RMA 和支持数据 |
信任信号高度分品类,公开资料大多缺失。
[CE018, CE019, CE020, CE027, CE034]| 日期 / 阶段 | 功能 / 里程碑 | 状态 | 含义 | 来源 |
|---|---|---|---|---|
| 2023 | 最新 UCIe 支持 | 已宣布 | 显示生态对齐 | 官方新闻稿 |
| 2024 | Samsung 代工平台移植 | 已宣布 | 支撑跨制造路径可移植性 | 官方新闻稿 |
| 2025 | 3nm PHY 达到 64Gbps | 已宣布 | 暗示先进节点技术成熟度 | 官方新闻稿 |
| 2025 | Arm Total Design 生态 | 已宣布 | 提升协同设计可信度 | 官方 + Arm |
| 2026 | NuLink-XD 发布 | 已宣布 | 延伸到光互连相邻用例 | 官方新闻稿 |
| 2026 | 商业化融资 | 已宣布 | 显示部署推进 | 官方 + 媒体 |
里程碑有公告支撑,但不应视为完整量产验证。
[CE006, CE007, CE008, CE009, CE015, CE024]5.4 信任、质量与技术结论
Eliyan 公开技术画像最弱的部分,不是有没有产品概念,而是缺少信任与质量披露。已检索来源没有提供支持 SLA、现场可靠性指标、正式缺陷数据、出口管制处理细节,或有助于投资人评估量产就绪度和质量纪律的产品级认证。在半导体基础设施里,信任不是消费者隐私页面;它体现为互操作证明、认证证据、支持成熟度,以及安全处理制造依赖的能力。Eliyan 通过标准参与和生态伙伴关系具备其中一些要素,但公开证据在路线图雄心上仍强于在质量体系披露上。最可防守的结论是,Eliyan 有一套技术上具体的架构故事,并有可信生态触点;但在其产品成熟度可被完全承销前,仍需要尽调支持、可靠性和合规。这个点重要,因为半导体客户常会运行认证项目,当供应商流程可见度弱时,即便技术看起来出色也可能被拒。Eliyan 因此不仅需要尽调性能主张,也需要尽调那些把实验室成功变成可靠部署的普通运营纪律。[CE018, CE019, CE020, CE027, CE030, CE031]
5.5 展示材料
06客户
6.1 客户分群与买方地图
Eliyan 的潜在客户不是普通企业,而是在 AI 计算、存储、封装和网络扩展上做设计决策的先进基础设施组织。公开材料反复指向 hyperscaler、AI 加速器开发商、存储供应商和基础设施厂商,说明这些才是相关分群。这意味着真实客户地图是多层的。用户可能是系统架构或封装团队,买方可能是平台或硅项目负责人,付费方则可能位于更大的数据中心或半导体预算内。这个点重要,因为互连产品通过技术设计项目采用,而不是通过轻量软件采购采用。也就是说,可寻址客户基数可能数量小,但单个账户价值大。Eliyan 看起来正对齐这种形态:一个供应商,成败取决于赢下少数战略重要账户,并满足很深的集成要求。实际来看,这会形成一种客户画像:哪怕只有少量活跃项目,也可能影响巨大。这也意味着,在公开收入披露出现很久以前,定性买方相关性就可能已经很强;这正是 Eliyan 当前材料中可见的模式。[CU001, CU002, CU012, CU018, CU030]
| 细分 | 买方 / 用户 / 付款方 | 主要用例 | 战略价值 | 缺口 |
|---|---|---|---|---|
| 超大规模云厂商 | 平台负责人 / 系统架构师 / 数据中心资本开支负责人 | 扩展 AI 计算和内存互连网络 | 潜在单户价值大 | 具名量产订单未公开 |
| AI 加速器厂商 | 芯片项目负责人 / 封装团队 / 产品预算 | 支撑 chiplet 和内存扩展 | 与核心产品直接契合 | 收入结构未知 |
| 内存供应商 | 内存和接口团队 / 平台 PM / 项目预算 | 解决内存墙瓶颈 | 重要技术邻近领域 | 无公开合同细节 |
| 基础设施和光互连厂商 | 网络或模块团队 / 系统用户 / 基础设施资本开支负责人 | 延伸至芯片到芯片和光互连网络 | 支撑扩张叙事 | 客户状态不清楚 |
分层来自官方市场和融资表述推断,不是客户名单披露。
[CU001, CU002, CU012, CU018, CU030]在少数成熟客户内部,Eliyan 可能从架构痛点切入,推进到验证、部署,再扩展到平台层面。
[CU011, CU016, CU018, CU024]6.2 具名证明与采用质量
Eliyan 最好的公开客户证据是间接的,但仍有意义。公司公开点名的战略投资人和生态参与者横跨 AMD、Arm、Meta、Cisco、Lumentum 以及光相邻交易对手;外部报道也反复把业务描述为与 hyperscale 和 AI 基础设施买家相关。这明显好过完全没有外部验证。但这不同于传统客户案例研究,后者会给出结果、合同范围、生产状态和续约历史。公开记录含有客户正在评估或部署 Eliyan 技术的措辞,但没有说明每个阶段有多少项目,也没有说明它们是否是付费量产合作。因此,本章把具名验证视为方向性正面,同时仍把客户证明质量评为不完整。Eliyan 有强生态拉力证据,但公开披露还不足以清楚区分投资人信念和持久客户收入。这个区分在 Eliyan 所处阶段尤其重要,因为 VC 和战略资本可以加速生态访问,但不能保证持续量产回款。一套扎实的客户证明包应把背书、试点、认证、部署和重复订单拆开。公开来源尚未给出这条阶梯。[CU003, CU004, CU005, CU006, CU007, CU008]
| 具名证明实体 | 细分 | 部署 / 用例 | 量产 / 试点 | 结果 / 局限 | 证据质量 |
|---|---|---|---|---|---|
| AMD 战略投资方阵营 | 计算生态 | 验证互连与先进计算路线图相关 | 未公开披露为量产客户 | 生态信号强,但采购深度未知 | 中 |
| Arm 生态参与 | 平台生态 | 支撑 chiplet 和 SoC 设计集成 | 是设计伙伴信号,不是采购证明 | 更能支撑可信度,不能支撑留存分析 | 中 |
| Cisco Investments 与 Lumentum | 网络与光互连生态 | 支撑向网络和光学领域扩展 | 公开记录里只是投资方验证 | 邻近领域证明不错,收入证明弱 | 中 |
| Meta / 超大规模云厂商群体提及 | 超大规模云厂商生态 | 显示大型 AI 基础设施利益方有兴趣 | 公开状态不清楚 | 若确认为真实部署,重要性可能很高 | 低-中 |
公开具名部署披露很薄,因此本枚举表并不完整,也把投资方、合作伙伴和隐含客户证明线索放在一起。
[CU004, CU005, CU006, CU007, CU009, CU019]| 指标 / 阶段 | 公开信号 | 日期 | 置信度 | 含义 | 缺失分母 |
|---|---|---|---|---|---|
| 评估中项目 | Series C 轮材料中声称 | 2026-07-29 | 中 | 销售漏斗顶部有活跃兴趣 | 未披露项目数量 |
| 部署项目 | Series C 材料中声称 | 2026-07-29 | 中 | 表明至少已有部分后期阶段参与 | 未披露数量或收入转化 |
| 战略投资者覆盖面 | 计算与光学厂商均在列 | 2026-01 至 2026-07 | 中 | 生态拉力看起来较广 | 不等于客户数量 |
| 商业化势头 | 多次出现在媒体报道中 | 2026-07-29 | 中 | 支撑采用率上升的叙事 | 没有出货量数据 |
轨迹证据来自阶段进展,不是完整指标。
[CU007, CU008, CU017, CU022, CU031]公开证据披露了阶段名称,但没有披露各阶段数量。
指数值表示相对可见度,不代表实际客户数量;Eliyan 披露的是阶段措辞,不是分母。
[CU003, CU007, CU008, CU017, CU031]证明质量在生态重要性上最高,在留存可见度上最低。
[CU009, CU013, CU019, CU021, CU033]6.3 留存、扩展与集中度
投资人想了解的客户持久性信息,几乎全部没有公开披露。已检索来源没有给出客户数量、头部账户占比、合同期限、续约率、NRR、GRR 或流失。这留下一个熟悉的半导体模式:战略相关性高,但初始 design-in 之后业务可重复性可见度有限。可能形态是一本由大机会构成、但集中度很高的账,一个重大项目的重要性超过几十个小账户。如果账户质量极高,这会很有吸引力;但项目延迟时,也会制造收入波动和执行风险。Eliyan 看起来确有合理的 land-and-expand 路径,因为其技术可以横跨裸片到裸片、存储、芯片到芯片,并最终延伸到光相邻层。只是,公开记录没有量化这个漏斗中的推进。可防守的看法是,扩展潜力真实存在,但留存和集中度仍是一阶尽调缺口。这个缺失信息,是本章即便面对亮眼战略名单仍保持谨慎的核心原因。对基础设施组件供应商而言,技术上受尊敬的产品和财务上持久的客户基础之间,通常只有在重复项目数据可见后才算跨过去。换个说法,Eliyan 可能已经拥有有价值的账户,但投资人无法从公开证据判断这些账户是孤立实验、正在放大的 design-in,还是长期平台关系。这种不确定性应直接压低对近期收入持久性的信心。[CU010, CU014, CU015, CU016, CU017, CU020]
| 指标 | 公开披露值 | 置信度 | 含义 | 尽调要求 |
|---|---|---|---|---|
| NRR / GRR | null | 低 | 看不到扩张的持续性 | 要求按产品披露同期群留存 |
| 流失 / 丢失项目 | null | 低 | 无法直接检验下行集中度 | 要求提供赢单 / 输单及延期记录 |
| 合同期限 | null | 低 | 关系收入可见度 | 要求提供合同期限和里程碑节奏 |
| 重复采购行为 | null | 低 | 需要区分一次性赢单与平台化采用 | 要求提供复购或后续设计数据 |
这些空值是有意保留的,因为公开记录没有披露留存指标。
[CU010, CU020, CU029, CU033]6.4 展示材料
07风险
7.1 监管与法律风险
Eliyan 最可见的公开监管风险是出口管制外溢。2026 年,美国围绕 AI 半导体的政策再次收紧,而 Eliyan 正切入先进 AI 连接和半导体基础设施这个政策制定者越来越视为战略性的领域。公司公开材料和融资公告没有解释 Eliyan 如何对产品分类、哪些地理区域最重要,或光学路线图会如何改变合规负担。这并不意味着现在存在违规,但确实制造了有意义的尽调不确定性。法律表面同样很薄。OpenCorporates 确认了实体存在,但公司的私营状态意味着投资人不能依赖他们通常会从上市发行人那里看到的诉讼、风险因素或治理披露。已检索集合中没有出现公开执法或诉讼,但当披露边界本身很窄时,未浮现事件只是较弱证据。[CR001, CR002, CR003, CR004, CR005, CR031]
| 规则 / 案件 | 司法辖区 | 状态 | 发生概率 | 严重性 | 缓释措施 | 剩余敞口 | 尽调路径 |
|---|---|---|---|---|---|---|---|
| AI 半导体出口管制 | 美国 / 全球触达 | 政策持续收紧 | 中 | 高 | 可能有内部合规计划,但未披露 | 分类证据拿出来之前,敞口高 | 要求提供出口分类和受限地区政策 |
| 产品出口分类 | 美国 / 客户所在地 | 未公开披露 | 中 | 高 | 有控制措施后可能可控 | 中高 | 要求提供 ECCN / 分类备忘录 |
| 私营公司法律披露限制 | 美国私营公司 | 当前 | 高 | 中 | 第三方核查与法律顾问尽调 | 中 | 要求提供诉讼、IP 和治理声明函 |
| 未来 IP / 标准争议 | 全球半导体生态 | 未见公开争议 | 低中 | 中 | 参与标准制定可能有帮助 | 中 | 要求提供专利版图和自由实施分析 |
行按严重性,以及公开记录中最可验证的事项排序。
[CR001, CR003, CR004, CR005, CR029, CR031]剩余风险集中在客户集中、运营转化、出口管制和估值压力。
[CR001, CR006, CR011, CR016, CR018, CR023]7.2 运营与产品风险
Eliyan 所处的位置让运营风险天然偏高:它卡在 AI 基础设施栈中最难的层之一,也就是裸片、存储、封装和越来越多光学之间的交界。即便底层技术很强,认证延迟、互操作挑战和制造依赖也可能推迟或缩小商业赢单。关于 AI 基础设施复杂度的独立评论支持这种担忧。公开记录没有披露良率、缺陷率或现场可靠性指标,也没有足够具体地点名代工厂或 OSAT 依赖,无法消除供应链风险。光学扩展增加了必须同时就位的外部伙伴和准备因素。实际后果是,Eliyan 可以在技术上正确,却在商业上来晚。技术相关性和运营转化之间的这道缝,是公司的核心承销风险之一。[CR006, CR007, CR008, CR009, CR023, CR027]
| 失效模式 | 发生概率 | 严重性 | 缓释成熟度 | 剩余敞口 | 未解决缺口 |
|---|---|---|---|---|---|
| 认证延迟 | 中 | 高 | 部分成熟 | 高 | 需要按项目提供实际周期 |
| 封装 / 集成复杂度 | 中 | 高 | 部分成熟 | 高 | 需要晶圆厂和 OSAT 依赖图谱 |
| 良率或缺陷意外 | Unknown | 高 | 公开披露低 | 高 | 需要可靠性和缺陷数据 |
| 光学路线图延期 | 中 | 中高 | 部分成熟 | 中高 | 需要里程碑时间表和依赖项 |
| 安全 / 质量认证缺口 | Unknown | 中 | 公开披露低 | 中 | 需要 ISO / 质量体系证据 |
运营风险更多来自深层基础设施复杂度,而不是互联网规模的网络安全态势。
[CR006, CR007, CR008, CR009, CR023, CR027]多数头部风险最终都会传导到客户转化、收入节奏、融资和估值。
[CR002, CR007, CR012, CR015, CR016, CR024]7.3 伙伴、客户与财务风险
风险栈中的伙伴侧和客户侧紧密相连。Eliyan 受益于与 Arm、AMD、Cisco、Lumentum 以及光学相关交易对手的生态联系,但这些联系也凸显商业化可能依赖少数复杂项目。客户集中因此很可能存在,而这个类别中的集中度会很快传导为收入波动、错过里程碑和估值波动。公开证据仍太薄,无法排除漫长试点周期或停滞的设计评估。财务风险又叠加其上。Eliyan 的公开记录对已融资金额着墨很多,对收入、利润率、burn 和 runway 几乎不披露。这意味着公司今天可以看起来资金充足,但如果 design-win 转化慢于 $1B 估值所暗示的速度,未来仍会面临资本压力。最重要的可见风险不是今天的偿付能力,而是相对于未来预期的证据不足。[CR010, CR011, CR012, CR015, CR016, CR017]
| 依赖 | 对手方 | 角色 | 集中度 | 失败场景 | 严重性 | 缓释措施 | 剩余敞口 |
|---|---|---|---|---|---|---|---|
| 平台生态适配 | Arm | 设计与生态对齐 | 中 | 融入客户 SoC 的路径偏弱 | 中 | 维持标准和伙伴支持 | 中 |
| 计算生态验证 | AMD / AMD Ventures | 战略验证 | 中 | 未转化为实际项目 | 中 | 拿下参考设计和商业赢单 | 中 |
| 网络相邻赛道 | Cisco Investments | 渠道 / 生态信号 | 中 | 投资者关系没有转化为销售 | 中 | 建立直接管线证据 | 中 |
| 光学相邻赛道 | Lumentum / Coherent | 与光学生态相关 | 中 | 光学路线图依赖第三方就绪度 | 高 | 分阶段里程碑和伙伴承诺 | 高 |
伙伴有帮助,但也让公司依赖狭窄生态。
[CR009, CR010, CR024, CR025, CR035]| 角色 / 职能 | 依赖或缺口 | 发生概率 | 严重性 | 缓释措施 | 尽调路径 |
|---|---|---|---|---|---|
| 创始人 / 技术领导力 | 路线图可信度集中在领导团队身上 | 中 | 高 | 董事会和战略投资者可以提供帮助 | 要求提供继任安排和梯队深度 |
| 招聘 / 用人 | 商业化需要持续招聘专业人才 | 中 | 中 | 融资支持团队扩张 | 要求披露关键空缺岗位悬而未决时长 |
| 项目管理 | 跨职能集成负荷高 | 中 | 高 | 公开信息未知 | 要求提供 PMO 和认证治理 |
| 商业化执行 | 少数大客户就能决定势头成败 | 中 | 高 | 战略支持方可能打开渠道 | 要求提供管线转化历史 |
执行风险重要,因为小团队必须把深技术转化为可靠部署。
[CR018, CR019, CR030, CR037]Eliyan 商业化离不开标准、伙伴、晶圆厂和大客户这张紧密网络。
[CR009, CR010, CR014, CR023, CR030, CR032]7.4 缓释措施、终止标准与总体看法
Eliyan 不是一家明显承压或显著不合规的公司;它的问题在于几项重要但仍未完全解开的风险叠在一起。公开信息里最有力的缓释因素,是已筹集资本规模、生态验证质量,以及产品叙事的技术细节。最薄弱的是法律与监管细节、量产级披露、客户持续性证据,以及全面财务透明度。这样的组合说明,管理风险要靠明确的否决阈值,而不是泛泛乐观。如果出口管制实质限制目标市场,如果光互连路线图节点延后,如果一两个旗舰项目停滞,或者下一轮融资早于重复客户证明,投资逻辑会迅速变弱。反过来,量产部署、客户集中度披露和清晰的认证指标,可以一次性拆掉几项最高严重度担忧。换句话说,下行路径会累积。任何单项开放风险都不必单独恶化,投资理由也会变弱;一次温和的客户延迟、一次温和的路线图滑坡,再叠加监管清晰度不足,就可能影响下一轮融资或压缩战略选择。委员会讨论应以这种累积框架为锚。[CR018, CR019, CR020, CR021, CR024, CR025]
7.5 图表
08估值
8.1 正向逻辑与反向逻辑
Eliyan 的估值争论有一点异常清楚:公司在解决一个真实且有战略重要性的基础设施问题。反向逻辑同样清楚:公开证据远远落后于雄心。Eliyan 的牛市情境靠品类相关性、技术细节、生态验证,以及 AI 系统设计变化可能让互连专家获得超额价值。反向逻辑是,许多基础设施创业公司在市场验证其能否把技术认可转成持久收入之前,都会这样描述自己。公开证据没有否定上行空间,但也不允许投资人把当前价格视为理所当然。这是一类公司质量可能很高、但价格纪律仍然极其重要的案子。这里,证据质量权重很大。[CV004, CV005, CV006, CV007, CV008, CV009]
| 建议 | 置信度 | 风险评级 | 估值立场 | 决策含义 |
|---|---|---|---|---|
| 继续研究 | 中 | 高 | 偏高 | 不要只凭公开证据承销上行空间 |
建议来自价格与证据不匹配,而不是市场野心不足。
[CV013, CV014, CV017, CV031, CV032, CV033]| 论点 | 何种情况会改变判断 |
|---|---|
| 品类痛点真实,技术定位有差异 | 竞争对手若把产品商品化,或客户停留在评估阶段,则转负面 |
| 战略方和伙伴名称提供强生态验证 | 验证若转化为具名量产账户,则转正面 |
| 2026 年大额融资支撑执行能力 | 收入证明出现前若又融资,则转负面 |
| 光学路线图扩大了战略相关性 | 光学里程碑若延期,或依赖风险扩大,则转负面 |
论点都配有会改变判断的条件。
[CV004, CV006, CV007, CV011, CV022, CV023]建议来自一个取舍:战略重要性很强,但在可见的独角兽估值下,运营证明仍有限。
[CV004, CV008, CV013, CV017, CV031]8.2 当前估值背景与进入纪律
最近的定价锚点是 2026 年 7 月 Series C 轮,估值 $1B;同年早些时候公司还完成 $50M 战略融资。这是强市场信号,但也是新投资人承销时最大的约束。Eliyan 已不再是押注优秀技术团队的便宜期权;它是一家以独角兽价格融资、但公开经营披露有限的半导体基础设施公司。营收、毛利率和资金续航都没有公开,投资人缺少一个简单分母去对照可见估值。因此,进入纪律不可让步。应假设轮次价格已经嵌入了客户转化、生态杠杆和光互连路线图进展的实质预期。举证责任于是落在能够证明估值合理、或揭示估值跑在商业现实前面的证据上。[CV001, CV002, CV003, CV010, CV013, CV014]
| 指标 | 公开披露值 | 置信度 | 含义 | 局限 |
|---|---|---|---|---|
| Series C 估值 | 1,000 USD M | 高 | 设定当前价格锚 | 没有收入分母 |
| Series C 规模 | 145 USD M | 高 | 显示投资人兴趣和资金续航支撑 | 轮次条款未披露 |
| 战略融资规模 | 50 USD M | 中 | 显示生态背书带来的动能 | 时点抬高执行预期 |
| 公开收入 / ARR | null | 低 | 传统价格-规模评估跑不通 | 需要管理层披露 |
本表把可见资本与缺失的经营指标拆开。
[CV001, CV002, CV003, CV008, CV029, CV030]信心最受客户转化、毛利质量和下一次融资时点影响。
[CV021, CV022, CV028, CV033, CV039]情景区间很宽,因为公开记录把价格讲清楚了,却没有讲清运营基数。
这些区间是挂钩牛市 / 基准 / 熊市情景的定性投资测算锚点,不是市场报价。
[CV001, CV010, CV018, CV019, CV034, CV035]8.3 情境与可比框架
公开网络证据不足以对可比公司做硬性归一,最稳妥的估值方法是情境分析,而不是一张伪精确的倍数表。上行情境下,如果 Eliyan 通过反复量产部署,证明自己是跨裸片间、内存和近光互连的关键连接层,当前或更高估值可以站得住。基准情境更保守:公司保持技术差异化和战略相关性,但投资人要从当前估值承销有意义上行,还需要更多客户、毛利和集中度证明。下行情境则在认证周期拖长、集中度偏高、且融资早于持久收入公开证据时出现。可比公司仍有方向意义,因为上市基础设施公司和规模化互连供应商展示了成熟披露应有的样子;但当前记录太薄,无法可靠输出倍数驱动的估值。因此,可比表有意保持克制:它只给出参考点和披露标准,不假装本流程已经提取出真正投资委员会备忘录所需的归一化市场数据。[CV011, CV012, CV016, CV018, CV019, CV024]
| 情景 | 假设 | 估值 / 回报逻辑 | 关键风险 | 概率信号 |
|---|---|---|---|---|
| 乐观 | 多个量产项目落地;光学路线图跑通;利润率质量显现 | 当前估值标记仍有明显上行空间 | 执行仍难,但价值捕获范围扩大 | 低-中 |
| 基准 | 技术相关性强,但客户和利润率证据不完整 | 当前估值在进一步尽调前大致合理到偏高 | 证据兑现慢 | 中 |
| 悲观 | 认证拖慢;集中度高;下一轮融资先于证据到来 | 当前估值标记大幅压缩 | 融资与捆绑风险 | 中 |
情景对里程碑很敏感,因为公开财务可比数据不完整。
[CV011, CV012, CV018, CV019, CV034, CV035]| 可比对象 | 指标 | 倍数 / 估值 / 状态 | 参考意义 | 局限 |
|---|---|---|---|---|
| Eliyan Series C 轮 | 私募轮 | 1.0B 估值 | 当前价格的直接锚点 | 不是经营倍数 |
| Astera Labs | 公开可比公司 | 申报主体 / 规模化互连背景 | 类别暴露更接近 | 本次公开网页检索未取到干净的标准化倍数 |
| Marvell | 公开可比公司 | 申报主体 / 更宽的基础设施平台 | 显示打包能力参照 | 业务组合宽得多 |
| AMD / Cisco | 公开可比公司 | 申报主体 / 生态平台 | 展示成熟披露该是什么样 | 范围过宽,不能直接给 Eliyan 定价 |
这一列举只给方向,不是完整倍数可比组,因为本流程没有提取可用的标准化公开指标。
[CV016, CV024, CV025]Eliyan 在市场重要性和技术差异化上得分高,但公开证明和估值支撑较弱。
[CV005, CV006, CV007, CV015, CV016, CV030]8.4 最终判断与尽调路径
仅凭当前公开证据,Eliyan 更像一个高质量的继续研究或跟踪标的,而不是现在就买入的高确信投资。这不是对公司的负面判决;只是承认可见价格很具体,而可见经营证明还不具体。如果管理层拿出反复量产部署、集中度数据、毛利质量,以及能支撑里程碑交付且没有近期压力的资金续航,结论可以上调。如果客户证明在下一轮融资周期前仍然间接,或市场围绕更宽产品包整合时执行拖慢,结论可以下调。因此,这是一条对证据敏感、带明确上调和下调触发条件的建议,而不是对团队质量或技术雄心的泛泛打分。用委员会语言说,Eliyan 值得再收集一轮数据,而不是立即否决。公司也许能证明其价格合理,但当前公开记录还没有做到。价格不应跑在证据前面。[CV017, CV020, CV022, CV023, CV026, CV028]
8.5 图表
免责声明
本报告是基于公开证据的尽调快照,不构成投资建议。重要财务、法律、技术及合同事实仍未公开;任何投资决定前,都应直接向管理层和原始文件核验。
证据索引
| 编号 | 陈述 | 可信度 | 来源 |
|---|---|---|---|
| CO001 | Eliyan is a Santa Clara, California-based semiconductor company focused on advanced connectivity for AI, cloud, and high-performance computing systems. | 高 | SO006, SO011 |
| CO002 | Third-party company databases reviewed for this run describe Eliyan as founded in 2021. | 中 | SO020 |
| CO003 | Eliyan’s current product scope includes NuLink PHY IP and NuGear chiplet families across compute, memory, and interconnect use cases. | 高 | SO004, SO006 |
| CO004 | Eliyan publicly positions its platform as spanning die-to-die, chip-to-chip, and emerging rack-to-rack electro-optical connectivity. | 高 | SO006, SO018 |
| CO005 | NuLink supports industry-standard BoW and UCIe connectivity as well as proprietary UMI and simultaneous bidirectional signaling modes. | 高 | SO004, SO005, SO019 |
| CO006 | On Eliyan’s products page, standard-package NuLink offerings are described for 100um to 130um bump pitches with 64 data lanes and up to 64 Gbps lane rates. | 中 | SO004 |
| CO007 | Eliyan’s advanced-package NuLink-AP offerings are described for 40um to 55um bump pitches and higher bandwidth-density metrics than standard-package implementations. | 中 | SO004 |
| CO008 | Eliyan argues that its architecture can extend chiplet economics from package interconnect into broader optical and system-level fabrics. | 高 | SO006, SO012 |
| CO009 | On July 29, 2026 Eliyan announced a $145 million Series C at a $1 billion valuation. | 高 | SO006, SO007, SO009 |
| CO010 | Seligman Ventures led Eliyan’s July 2026 Series C and Cisco Investments and Lumentum were named as new strategic investors. | 高 | SO006, SO007, SO009 |
| CO011 | Umesh Padval joined Eliyan’s board as part of the July 2026 financing. | 高 | SO006, SO009 |
| CO012 | Public 2026 materials identify Ramin Farjadrad, Patrick Soheili, and Syrus Ziai as Eliyan’s repeat-founder core. | 高 | SO006, SO016 |
| CO013 | Ramin Farjadrad is Eliyan’s co-founder and CEO. | 高 | SO003, SO016 |
| CO014 | Patrick Soheili is an Eliyan co-founder and the company’s public business-strategy lead. | 高 | SO003, SO016 |
| CO015 | Syrus Ziai is a public Eliyan co-founder and senior operations or engineering leader. | 高 | SO003, SO016 |
| CO016 | Ramin Farjadrad’s publicly cited background includes Aquantia and Velio plus standards work around Ethernet and chiplet interconnects. | 高 | SO016, SO021, SO011 |
| CO017 | Patrick Soheili’s published background includes executive or commercialization roles across eSilicon, Altera, AMD, and TRW. | 中 | SO016 |
| CO018 | Syrus Ziai’s published background includes engineering leadership roles at Nuvia, Ikanos, Qualcomm, and PsiQuantum. | 中 | SO016 |
| CO019 | Jason Taylor joined Eliyan’s board in November 2023. | 中 | SO014 |
| CO020 | Eliyan announced a $40 million Series A in November 2022 led by Tracker Capital with participation from Celesta, Intel Capital, and Micron. | 中 | SO011 |
| CO021 | Eliyan announced a $60 million Series B in March 2024 co-led by Samsung Catalyst Fund and Tiger Global with participation from Intel Capital, SK hynix, and others. | 中 | SO010 |
| CO022 | Eliyan announced a separate $50 million strategic financing in January 2026 from AMD, Arm, Coherent, Meta, Samsung Catalyst Fund, and Intel Capital. | 高 | SO012, SO013 |
| CO023 | Summing Eliyan’s publicly disclosed financings since 2022 implies at least $295 million of announced capital. | 中 | SO006, SO010, SO011, SO012 |
| CO024 | The July 2026 Series C release says Eliyan had more than 100 patents, a marquee customer base, and validation from 12 leading strategic investors. | 中 | SO006 |
| CO025 | Eliyan says leading hyperscale, AI accelerator, memory, and infrastructure customers are evaluating or deploying its technologies. | 中 | SO006 |
| CO026 | Eliyan’s November 2022 Series A release paired financing with a successful 5nm tapeout and expected first silicon in Q1 2023. | 中 | SO011 |
| CO027 | Eliyan’s March 2024 Series B release said NuLink had taped out on TSMC 3nm targeting up to 64 Gbps per link. | 中 | SO010 |
| CO028 | In October 2024 Eliyan announced a 64Gbps 3nm NuLink-2.0 milestone on standard organic packaging. | 中 | SO018 |
| CO029 | In November 2024 Eliyan announced a Samsung SF4X port positioned for custom HBM4 base-die and disaggregated-AI use cases. | 中 | SO017 |
| CO030 | In November 2025 Eliyan said it joined Arm’s Total Design ecosystem and would support Arm’s Foundation Chiplet System Architecture. | 中 | SO015 |
| CO031 | In July 2026 Eliyan launched NuLink-XD, a 224G PAM4 SerDes family aimed at chip-to-chip and chip-to-module links for AI scale-up. | 中 | SO018 |
| CO032 | Eliyan’s technology pages claim standard-package implementations can avoid advanced packaging cost, complexity, and supply-chain constraints while preserving high bandwidth and low power. | 高 | SO005, SO011 |
| CO033 | Eliyan’s public team page shows a broader management bench including finance, legal, sales, memory-products, and interface-chiplet leadership roles. | 中 | SO003 |
| CO034 | Tracxn lists Eliyan as a Series C company with 97 employees as of June 2026. | 低 | SO020 |
| CO035 | Reuters reported that Farjadrad described current AI-chip utilization as only about 30% to 40% because chips cannot be fed data fast enough. | 中 | SO007 |
| CO036 | Electronic Design coverage tied Eliyan’s memory-interconnect work to HBM4, SPHBM4, and HBM5-era bandwidth constraints in AI and HPC systems. | 高 | SO022, SO012 |
| CO037 | EE Times argues the chiplet market is entering a consolidation phase and still faces unresolved challenges around IP integration, testing, security, and optical transition. | 中 | SO023 |
| CO038 | SemiEngineering describes AI infrastructure as being at a crossroads where cost, complexity, and deployment constraints can slow the translation of technical potential into production systems. | 中 | SO024 |
| CO039 | Based on Eliyan’s product pages and strategic-funding description, the commercial model appears to blend PHY-IP licensing with chiplet product sales to chipmakers and system builders. | 中 | SO004, SO012 |
| CO040 | Reviewed public materials do not disclose audited revenue, ARR, customer concentration, full ownership percentages, or committee-level governance details. | 中 | SO006, SO003, SO020 |
| CM001 | Eliyan’s direct market sits in the connectivity layer for multi-die AI systems rather than in the entire semiconductor market. | 高 | SM001, SM002, SM003 |
| CM002 | The most relevant public category for Eliyan is chiplet interconnect and UCIe-style connectivity. | 高 | SM001, SM006, SM010 |
| CM003 | Status-quo substitutes for Eliyan include proprietary die-to-die links, advanced-packaging-heavy architectures, rack-scale switches, and photonic interconnect systems. | 高 | SM017, SM019, SM020, SM021, SM023 |
| CM004 | Eliyan explicitly targets AI compute, memory, and networking systems rather than a single point product niche. | 高 | SM003, SM004 |
| CM005 | Fortune Business Insights values the chiplet interconnect and UCIe ecosystem market at USD 3.28 billion in 2026. | 中 | SM006 |
| CM006 | Fortune Business Insights projects the chiplet interconnect and UCIe ecosystem market to reach USD 23.47 billion by 2034. | 中 | SM006 |
| CM007 | GM Insights values the global chiplet interconnect market at USD 2.89 billion in 2026. | 中 | SM007 |
| CM008 | Intel Market Research values the broader chiplet market at USD 7.31 billion in 2026. | 低 | SM008 |
| CM009 | PatSnap places the advanced-packaging market at roughly USD 49-55 billion by 2026. | 中 | SM009 |
| CM010 | PatSnap describes 2026 as an inflection point in which UCIe has become the de facto open standard for die-to-die communication. | 高 | SM009, SM010 |
| CM011 | Fortune says UCIe is expected to be the fastest-growing subsegment within the chiplet interconnect ecosystem. | 中 | SM006 |
| CM012 | Fortune identifies AI and machine-learning accelerators as the largest end segment in the chiplet interconnect and UCIe ecosystem market. | 中 | SM006 |
| CM013 | Fortune says North America holds the largest current share of the chiplet interconnect and UCIe ecosystem market. | 中 | SM006 |
| CM014 | These public TAM lenses are nested rather than directly comparable because they measure interconnect, broader chiplets, and packaging at different levels of the stack. | 中 | SM006, SM007, SM008, SM009 |
| CM015 | A single headline TAM would overstate Eliyan’s direct SAM because Eliyan monetizes connectivity layers rather than full packaging and full-system budgets. | 高 | SM001, SM002, SM006, SM009 |
| CM016 | Likely direct buyers for Eliyan-like interconnects are AI accelerator and silicon-platform design teams. | 中 | SM001, SM004, SM017 |
| CM017 | The operational users are system architects, memory architects, packaging teams, and infrastructure engineers trying to relieve bandwidth and latency bottlenecks. | 中 | SM002, SM011, SM012 |
| CM018 | The ultimate payer is often a hyperscaler or OEM platform budget even when the immediate integration decision sits with an ASIC or module team. | 中 | SM004, SM017, SM019 |
| CM019 | Astera Labs markets rack-scale AI connectivity and describes the rack as the new unit of compute. | 高 | SM017, SM018 |
| CM020 | Broadcom frames co-packaged optics around bandwidth density and power efficiency for AI infrastructure. | 中 | SM019 |
| CM021 | Lightmatter markets Passage and Guide photonic interconnect components as a complete interconnect platform for AI scale. | 中 | SM020 |
| CM022 | Ayar Labs markets co-packaged optics for connecting thousands of GPUs across racks into unified AI clusters. | 中 | SM021 |
| CM023 | Celestial AI’s current positioning emphasizes connectivity as a critical layer in modern AI data center infrastructure. | 中 | SM022 |
| CM024 | Marvell and Lumentum’s OFC 2026 optical circuit-switching demonstration shows major vendors are investing in low-latency optical scale-up fabrics for hyperscaler AI. | 中 | SM023 |
| CM025 | Eliyan’s own materials and Electronic Design both frame memory and I/O bottlenecks as core drivers of demand for better interconnect. | 高 | SM001, SM004, SM011 |
| CM026 | SemiAnalysis says AI systems require memory with higher capacity, lower latency, higher bandwidth, and improved energy efficiency as models grow in complexity. | 中 | SM012 |
| CM027 | Eliyan claims that compute capability has risen far faster than DRAM bandwidth and I/O bandwidth, producing memory and I/O walls. | 高 | SM001, SM002 |
| CM028 | Eliyan’s January 2026 strategic financing release says its NuGear chiplets target 1.6T to 12.8Tbps link bandwidths for demanding AI accelerator and memory architectures. | 中 | SM004 |
| CM029 | The UCIe Consortium says UCIe 3.0 doubles planar bandwidth density and adds runtime power and manageability enhancements for AI and HPC systems. | 中 | SM010 |
| CM030 | Fortune says the UCIe Consortium included more than 130 member companies in April 2026. | 中 | SM006 |
| CM031 | Open-standards momentum lowers integration friction for the kind of multi-vendor chiplet ecosystem Eliyan wants to serve. | 高 | SM006, SM010, SM024 |
| CM032 | Eliyan’s strategic investor roster implies that compute, hyperscaler, memory, networking, and optical players see interconnect as a critical architectural layer. | 高 | SM003, SM004 |
| CM033 | SemiEngineering says chiplet-based designs are becoming more complex and present a growing array of architectural tradeoffs. | 中 | SM013 |
| CM034 | SemiEngineering’s memory-wall coverage reinforces that bandwidth scaling is constrained by routing density, packaging, and system complexity. | 高 | SM014, SM012 |
| CM035 | EE Times argues the chiplet market faces unresolved issues around IP integration, testing, security, and the transition toward optical I/O. | 中 | SM016 |
| CM036 | Adjacent vendors such as Astera, Broadcom, Lightmatter, Ayar, Celestial AI, and Marvell-linked optical systems compete for the same connectivity budget from different layers of the stack. | 高 | SM017, SM019, SM020, SM021, SM022, SM023 |
| CM037 | Public evidence supports only a broad product-near market envelope for Eliyan, not a precise bottom-up SOM. | 高 | SM006, SM007, SM008, SM009, SM003 |
| CM038 | Current public materials do not disclose Eliyan pricing, attach rate, content per design win, or named customer count, which prevents a bottom-up share model. | 高 | SM003, SM004, SM005 |
| CM039 | The strongest adoption debate is not whether bandwidth demand exists but whether connectivity value is captured by neutral chiplet specialists or by larger adjacent incumbents. | 中 | SM015, SM016, SM023 |
| CM040 | The market’s decision chain is multi-stage, with workload needs flowing through silicon, packaging, and rack/network architecture before converting into deployment. | 高 | SM017, SM019, SM023 |
| CP001 | Eliyan positions its offering around PHY IP, chiplets, and memory/interconnect components rather than full accelerators. | 高 | SP001, SP002 |
| CP002 | Astera Labs sells connectivity silicon and board-level fabric components that compete for AI system I/O budget. | 高 | SP006, SP007 |
| CP003 | Broadcom competes from an incumbent networking and optics base rather than a chiplet-only starting point. | 中 | SP008, SP009 |
| CP004 | Lightmatter markets photonic interconnect infrastructure aimed at AI cluster scaling. | 中 | SP010 |
| CP005 | Celestial AI markets a photonic fabric approach for memory and AI connectivity. | 中 | SP011 |
| CP006 | Ayar Labs is an adjacent optical I/O competitor focused on moving data off-package with photonics. | 中 | SP012 |
| CP007 | Marvell is pairing optical and custom-silicon assets with NVIDIA ecosystem alignment, which raises the competitive bar for standalone startups. | 中 | SP013, SP014 |
| CP008 | Alphawave is commercializing UCIe IP at advanced nodes, showing that IP-first rivals are also moving quickly up the performance curve. | 中 | SP015, SP016 |
| CP009 | The UCIe ecosystem is broadening, making standards compliance less differentiating over time. | 中 | SP016, SP017, SP018 |
| CP010 | Eliyan’s differentiation claim depends on higher shoreline bandwidth, simultaneous bidirectional signaling, and memory-oriented links rather than standard compliance alone. | 中 | SP002, SP024 |
| CP011 | Competitor classes span direct D2D/UCIe IP vendors, retimer/fabric incumbents, optical interconnect vendors, and internal build by hyperscalers. | 中 | SP006, SP009, SP014, SP016 |
| CP012 | Buyer evaluation is likely split between package-level die-to-die needs and rack-scale optical/network scaling needs. | 中 | SP003, SP013, SP014, SP019 |
| CP013 | Eliyan is earlier in scale than public incumbents but may be more focused on the specific interconnect bottlenecks of chiplet AI systems. | 中 | SP005, SP006, SP013 |
| CP014 | Competitors with broader platforms can bundle connectivity into larger silicon or networking programs, increasing price pressure. | 中 | SP007, SP009, SP014 |
| CP015 | Where pricing is not public, packaging and integration support are a meaningful source of switching cost. | 中 | SP016, SP021 |
| CP016 | Internal build remains a real substitute for top hyperscalers and custom-silicon programs. | 中 | SP014, SP017 |
| CP017 | Eliyan’s strategic financing from hyperscalers and infrastructure partners modestly improves access relative to startups without design-partner backing. | 中 | SP004, SP025 |
| CP018 | Competitors tied to large installed ecosystems may have stronger channel leverage than Eliyan in near-term procurement. | 中 | SP006, SP009, SP014 |
| CP019 | Optical interconnect entrants increase the risk that part of the value pool shifts beyond the package, away from pure electrical D2D differentiation. | 中 | SP010, SP011, SP012, SP013 |
| CP020 | Packaging complexity and standards churn can slow customer switching even when a rival offers better raw link metrics. | 中 | SP016, SP019, SP021 |
| CP021 | The public evidence does not support a comprehensive apples-to-apples pricing comparison across the peer set. | 中 | SP006, SP007, SP010 |
| CP022 | The public evidence also does not prove broad production wins for most private optical competitors, limiting certainty on adoption ranking. | 中 | SP010, SP011, SP012 |
| CP023 | The strategic question for Eliyan is whether customers value interoperable IP and chiplets more than vendor-specific full-stack bundles. | 中 | SP001, SP014, SP016 |
| CP024 | Public-company incumbents can tolerate longer commercialization cycles than venture-backed specialists. | 中 | SP013, SP014, SP022 |
| CP025 | Consolidation risk is material because chiplet infrastructure categories may narrow to a few scaled winners. | 中 | SP022, SP017 |
| CP026 | Eliyan’s move into electro-optical interconnects is partly a response to adjacent competitors pushing connectivity beyond copper. | 中 | SP005, SP013, SP014 |
| CP027 | Astera and incumbent ecosystem vendors likely win where customers prefer validated system platforms over novel interface changes. | 中 | SP006, SP007, SP021 |
| CP028 | Eliyan may win where customers need custom memory and die-to-die bandwidth density before broader standards roadmaps catch up. | 中 | SP002, SP020, SP024 |
| CP029 | Standards participation is necessary but insufficient because competitors are also demonstrating interoperability milestones. | 中 | SP015, SP016, SP025 |
| CP030 | The strongest public competitor proof exists for established vendors and consortia, not for Eliyan’s exact private-company peer set. | 中 | SP016, SP022 |
| CP031 | Large ecosystem partners such as Arm can reinforce Eliyan’s credibility but do not remove head-to-head pressure from well-funded rivals. | 中 | SP025, SP005 |
| CP032 | Memory-wall and bandwidth constraints create room for multiple approaches, including electrical D2D, optical links, and broader platform integration. | 中 | SP019, SP023, SP024 |
| CP033 | Because Eliyan targets infrastructure rather than end-user silicon, competitive outcomes may hinge on design-in timing more than brand awareness. | 中 | SP003, SP017, SP021 |
| CP034 | Optics-oriented competitors could be collaborators in one program and substitutes in another, implying a fluid coopetition dynamic. | 中 | SP005, SP013, SP014 |
| CP035 | The public record supports a capability comparison better than a revenue-share comparison. | 中 | SP017, SP018, SP022 |
| CP036 | Hyperscaler internal build and incumbent bundle power are the clearest displacement threats to Eliyan’s standalone value capture. | 中 | SP014, SP017, SP022 |
| CP037 | Eliyan’s competitive moat is currently more technology- and partner-access-driven than customer-lock-in-driven. | 中 | SP004, SP020, SP021 |
| CP038 | The absence of public customer reference detail makes durability harder to prove than raw technical ambition. | 中 | SP005, SP020 |
| CP039 | Eliyan curates multiple third-party technical and funding mentions on its in-the-news page, indicating an active effort to shape category visibility. | 低 | SP026 |
| CP040 | Astera Labs maintains an active public news surface that reinforces its category visibility with infrastructure buyers. | 低 | SP027 |
| CI001 | Eliyan’s public monetization appears to span licensing or supplying interconnect IP, PHY technology, and chiplet/connectivity products rather than end-user AI software. | 高 | SI001, SI002 |
| CI002 | The company’s product messaging centers on solving memory-wall and connectivity bottlenecks for AI and HPC system builders. | 高 | SI001, SI002 |
| CI003 | Eliyan does not publicly disclose revenue, ARR, gross margin, or unit shipments in the retrieved 2026 funding materials. | 高 | SI004, SI009, SI010 |
| CI004 | The January 2026 strategic financing and July 2026 Series C indicate capital raising remains a core part of the operating plan as commercialization scales. | 高 | SI003, SI004, SI017 |
| CI005 | Strategic financing was framed around commercialization of NuLink PHY and NuGear chiplets, linking fundraising directly to product rollout. | 中 | SI003, SI017, SI018 |
| CI006 | The July 2026 Series C was framed around advancing electro-optical interconnects for AI infrastructure. | 中 | SI004, SI009, SI024 |
| CI007 | The public record supports a multi-stream hardware/IP business model, but not the current mix between licensing, silicon, services, or support. | 中 | SI002, SI017 |
| CI008 | Because realized pricing is undisclosed, public sources do not support a clean estimate of ASPs, royalty rates, or gross profit per design win. | 中 | SI002, SI004 |
| CI009 | Eliyan’s sales motion likely involves long design cycles and technical qualification rather than transactional self-serve demand. | 中 | SI002, SI027 |
| CI010 | Strategic investors from hyperscalers and infrastructure providers may shorten some commercial cycles by increasing ecosystem access. | 中 | SI003, SI025 |
| CI011 | The company’s funding history now includes at least a $40M Series A, $60M Series B, $50M strategic financing, and $145M Series C. | 中 | SI006, SI005, SI003, SI004 |
| CI012 | Those disclosed rounds imply at least $295M of announced capital, excluding any undisclosed seed or secondary transactions. | 中 | SI006, SI005, SI003, SI004 |
| CI013 | A $1B valuation at Series C suggests investors are underwriting future design-win and market-share potential rather than disclosed current revenue. | 中 | SI004, SI010 |
| CI014 | Open roles on the careers page signal ongoing operating-expense commitments across engineering and commercialization. | 中 | SI007 |
| CI015 | Hardware interconnect programs likely require expensive foundry, packaging, validation, and support work that pure-software businesses do not bear. | 中 | SI002, SI027 |
| CI016 | Electro-optical expansion can add R&D and partner-integration cost before associated revenue fully materializes. | 中 | SI004, SI024 |
| CI017 | The public evidence does not disclose inventory balances, purchase commitments, or working-capital needs. | 中 | SI004, SI015 |
| CI018 | The public evidence also does not disclose debt facilities, project finance, or customer prepayments. | 中 | SI004, SI019 |
| CI019 | Strategic investors from AMD, Arm, and Meta in early 2026 validate interest but do not substitute for recurring customer receipts. | 中 | SI025, SI017 |
| CI020 | Comparable public infrastructure vendors disclose through SEC filing regimes, while Eliyan remains far less transparent as a private company. | 中 | SI020, SI021, SI022, SI023 |
| CI021 | That disclosure gap limits any precise public estimate of burn, runway, or gross-margin trajectory. | 中 | SI020, SI021, SI022, SI023 |
| CI022 | Trade-press commentary on chiplet complexity supports the view that commercialization requires more capital and longer validation than simple IP stories imply. | 中 | SI026, SI027 |
| CI023 | The company is likely pre-scale or early-scale on revenue relative to its valuation because no revenue milestones are highlighted in funding coverage. | 中 | SI009, SI010, SI013 |
| CI024 | Private-company data is sufficient to judge funding momentum, but insufficient to judge revenue quality. | 中 | SI004, SI015, SI016 |
| CI025 | Eliyan’s forward financial thesis depends on converting technical relevance into repeat design wins before another large financing is required. | 中 | SI004, SI026 |
| CI026 | The business likely has a small number of high-value customer programs rather than a broad base of low-ticket accounts. | 中 | SI002, SI003 |
| CI027 | Any customer concentration could create lumpy revenue recognition and milestone-based cash flow, but public evidence is missing. | 中 | SI003, SI004 |
| CI028 | The official materials emphasize product performance and ecosystem adoption more than near-term monetization metrics. | 中 | SI001, SI004 |
| CI029 | Series C timing only six months after the strategic financing suggests an aggressive capital plan aligned with fast-moving AI infrastructure demand. | 中 | SI003, SI004 |
| CI030 | Third-party funding articles consistently describe commercialization and manufacturing expansion, reinforcing that the capital is for scale-up rather than rescue financing. | 中 | SI012, SI013, SI024 |
| CI031 | No public source retrieved provides CAC, payback, or sales-efficiency ratios, so those must be treated as diligence asks rather than inferred. | 中 | SI004, SI015 |
| CI032 | Likewise, no public source retrieved provides gross-margin disclosure, making hardware-versus-IP mix especially important in diligence. | 中 | SI004, SI017 |
| CI033 | The 2026 funding stack reduces immediate solvency concern but increases pressure to prove commercialization at a valuation that already assumes success. | 中 | SI004, SI010, SI011 |
| CI034 | If the company’s model leans toward custom engagements, revenue recognition could be milestone-driven and uneven. | 低 | SI002, SI017 |
| CI035 | If the model leans toward reusable IP and chiplets, gross margin could improve with scale, but that is not yet publicly evidenced. | 低 | SI002, SI024 |
| CI036 | The best-supported financial conclusion is that Eliyan is well financed relative to disclosed milestones but still under-disclosed for full underwriting. | 中 | SI004, SI010, SI015 |
| CI037 | Because the company is private, historical round chronology is visible but current cash-on-hand remains undisclosed. | 中 | SI014, SI015 |
| CI038 | Entity-registry visibility and media coverage confirm company existence and momentum but add little evidence on core unit economics. | 中 | SI019, SI008 |
| CE001 | Eliyan’s product surface spans connectivity IP, PHY technology, chiplets, and memory-interconnect elements rather than full compute silicon. | 高 | SE001, SE002 |
| CE002 | The product is sold into AI, HPC, cloud, and memory-centric system design workflows. | 高 | SE001, SE003 |
| CE003 | Public materials center on NuLink, NuGear, and UMI-style building blocks for multi-die and memory-wall problems. | 中 | SE001, SE007, SE012 |
| CE004 | Eliyan’s architecture story emphasizes chiplet interconnect bandwidth density and memory scaling rather than monolithic die expansion. | 中 | SE002, SE007, SE016 |
| CE005 | The company’s simultaneous bi-directional signaling and packaging white papers indicate a technical effort to differentiate below the marketing layer. | 高 | SE005, SE006 |
| CE006 | The 64Gbps-in-3nm announcement suggests the product is beyond concept stage and tied to advanced-node implementation work. | 中 | SE008, SE017 |
| CE007 | NuLink-XD extends the roadmap from package-level links toward chip-to-chip and chip-to-module optical-adjacent connectivity. | 中 | SE009, SE025 |
| CE008 | UCIe support and Arm ecosystem participation indicate Eliyan is trying to fit into a broader interoperability stack, not only proprietary deployments. | 中 | SE010, SE013, SE014 |
| CE009 | Samsung foundry porting suggests process portability matters to the go-to-market strategy. | 中 | SE011 |
| CE010 | The recruiting surface is the clearest public developer-signal proxy: Eliyan is still staffing technical functions, implying active product development rather than maintenance mode. | 中 | SE015 |
| CE011 | The core user appears to be the system-architecture or packaging team that needs better die-to-die, memory, or optical-adjacent throughput. | 中 | SE003, SE016 |
| CE012 | Public evidence supports architecture intent, but not a complete SKU list or public datasheet set for every claimed module. | 中 | SE001, SE004 |
| CE013 | The product workflow begins with system bottlenecks, then interface choice, then packaging and interoperability decisions before production integration. | 中 | SE002, SE005, SE019 |
| CE014 | Memory-wall positioning is central to the technical thesis and differentiates Eliyan from generic connectivity vendors. | 中 | SE007, SE016, SE020 |
| CE015 | Roadmap credibility improves when the same company shows white papers, foundry portability, standards support, and funding aimed at commercialization. | 中 | SE005, SE011, SE012 |
| CE016 | The optical roadmap raises dependency on external module, ecosystem, and packaging readiness beyond Eliyan’s own IP blocks. | 中 | SE009, SE022, SE024 |
| CE017 | Arm ecosystem entry matters because broader SoC and chiplet programs often require co-design with major CPU/platform partners. | 中 | SE013, SE014 |
| CE018 | Public evidence does not show customer-support SLAs, field-reliability metrics, or production defect rates. | 中 | SE001, SE004 |
| CE019 | Public evidence likewise does not confirm security certifications, functional-safety approvals, or formal quality-system certifications. | 中 | SE001, SE004 |
| CE020 | Because the company sells deep infrastructure components, trust signals are more about interoperability, qualification, and supply access than about end-user privacy policy. | 中 | SE010, SE013, SE019 |
| CE021 | Independent market commentary supports the premise that packaging and memory constraints are real, not invented by company marketing. | 中 | SE018, SE020 |
| CE022 | Independent commentary also reinforces that AI interconnect architectures are changing quickly, which can both help and threaten Eliyan’s roadmap. | 中 | SE021, SE023, SE024 |
| CE023 | The product appears technically ambitious but still under-documented publicly compared with mature semiconductor platforms. | 中 | SE004, SE015, SE017 |
| CE024 | Commercialization funding in January 2026 implies the company is transitioning from pure technology proof toward broader deployment readiness. | 中 | SE012, SE025 |
| CE025 | Public sources do not establish broad named production deployments, so product maturity should be treated as promising but not fully market-proven. | 中 | SE012, SE017 |
| CE026 | Eliyan’s most defensible moat today is architecture-specific know-how around D2D, memory, and packaging interactions. | 中 | SE005, SE006, SE007 |
| CE027 | The weakest public area is trust-and-quality disclosure, not the existence of an architecture story. | 中 | SE018, SE019 |
| CE028 | Optical-adjacent expansion may improve future TAM while also increasing implementation complexity and partner dependence. | 中 | SE009, SE024, SE025 |
| CE029 | Foundry portability and standards work imply Eliyan wants its technology to travel across customer roadmaps rather than remain locked to one manufacturing path. | 中 | SE010, SE011, SE014 |
| CE030 | The public developer-signal proxy is recruiting and ecosystem participation rather than open-source or package-registry activity, which is typical for semiconductor infrastructure companies. | 中 | SE014, SE015 |
| CE031 | Roadmap evidence is stronger than support-operation evidence in the retrieved public record. | 中 | SE008, SE009, SE018 |
| CE032 | Eliyan’s architecture narrative is easier to verify than its production readiness because public sources concentrate on design claims and partner news. | 中 | SE004, SE012, SE019 |
| CE033 | The company’s product-tech profile fits an enabling component supplier that must win co-design trust from a relatively small set of sophisticated buyers. | 中 | SE002, SE003, SE017 |
| CE034 | No retrieved source proves formal compliance coverage for export controls, security review, or quality certifications at the product level. | 中 | SE021, SE022 |
| CE035 | The public record is sufficient to map product logic and dependencies, but not to clear all reliability and support risks. | 中 | SE018, SE019, SE021 |
| CE036 | Eliyan maintains a broad press-release surface that documents multiple roadmap milestones rather than a single isolated launch. | 低 | SE027 |
| CE037 | Additional funding coverage from smaller outlets broadly matches the official roadmap narrative, even if it adds limited technical detail. | 低 | SE028, SE029, SE030 |
| CU001 | Eliyan targets sophisticated infrastructure buyers rather than broad enterprise end users. | 高 | SU001, SU002 |
| CU002 | The likely customer set includes hyperscalers, AI accelerator developers, memory suppliers, and infrastructure vendors. | 高 | SU001, SU004 |
| CU003 | Public evidence is stronger on strategic investor and ecosystem names than on named production customers. | 高 | SU003, SU004, SU005 |
| CU004 | AMD is named in third-party coverage of Eliyan’s strategic-investor cohort, indicating buyer-side ecosystem validation even if production use is not disclosed. | 中 | SU018, SU019, SU020 |
| CU005 | Arm is publicly tied to Eliyan through ecosystem participation, which is a meaningful design-partner signal but not proof of recurring purchase. | 中 | SU006, SU003 |
| CU006 | Cisco Investments and Lumentum joined the 2026 Series C, broadening Eliyan’s relevance into networking and optical ecosystems. | 中 | SU004, SU007, SU009 |
| CU007 | The Series C materials also claim commercial momentum across hyperscale, AI accelerator, memory, and infrastructure customers evaluating or deploying Eliyan technologies. | 中 | SU004, SU013 |
| CU008 | That claim is directionally encouraging, but the public record does not identify which of those evaluations are pilots versus production deployments. | 中 | SU004, SU015 |
| CU009 | Strategic investors are valuable customer-proof proxies because they validate technical fit, but they are weaker than signed multiyear purchase disclosures. | 中 | SU003, SU005, SU018 |
| CU010 | The public record does not disclose customer count, churn, NRR, GRR, renewal rate, or contract length. | 中 | SU004, SU024 |
| CU011 | Because the product sits deep in infrastructure design cycles, adoption likely advances through qualification, deployment, and then broader program expansion rather than fast seat-based growth. | 中 | SU002, SU022 |
| CU012 | Named ecosystem participants span compute, memory, networking, and optical-adjacent companies, implying Eliyan sells into a multi-party design chain. | 中 | SU003, SU004, SU009, SU011 |
| CU013 | The most credible public customer signal is strategic alignment from technically relevant counterparties, not visible retention metrics. | 中 | SU005, SU006, SU007 |
| CU014 | A concentrated customer base is likely because semiconductor interconnect suppliers typically win a small number of large programs first. | 中 | SU001, SU022 |
| CU015 | That likely concentration raises lumpiness risk if one hyperscaler, accelerator vendor, or memory program slips. | 中 | SU022, SU023 |
| CU016 | Public materials imply land-and-expand potential because connectivity can span die-to-die, memory, chip-to-chip, and rack-scale architectures. | 中 | SU002, SU004 |
| CU017 | Yet the public record does not quantify how many programs have progressed from evaluation to deployment. | 中 | SU004, SU015 |
| CU018 | The company’s likely buyer, user, and payer are different stakeholders inside advanced infrastructure accounts, which complicates procurement. | 中 | SU001, SU002 |
| CU019 | Strategic investors from AMD, Arm, Meta, Coherent, Cisco, and Lumentum would collectively suggest strong ecosystem relevance if diligence confirms their depth of engagement. | 中 | SU003, SU004, SU005, SU009, SU011 |
| CU020 | No public source retrieved proves durable repeat-purchase behavior or renewal cycles. | 中 | SU004, SU021 |
| CU021 | The absence of customer logos tied to measurable outcomes means the customer chapter remains proof-constrained even though interest appears real. | 中 | SU003, SU004, SU024 |
| CU022 | Independent coverage consistently frames Eliyan as a solution to critical AI bottlenecks, which supports buyer relevance even without granular deployment counts. | 中 | SU014, SU015, SU017 |
| CU023 | Procurement friction is likely high because adoption touches architecture, packaging, and systems teams rather than one software budget owner. | 中 | SU002, SU022 |
| CU024 | If Eliyan wins inside a platform, expansion could come from additional links, adjacent modules, or later optical layers rather than from simple seat growth. | 中 | SU002, SU004 |
| CU025 | The public evidence is sufficient to segment likely customers and identify named validation proxies, but insufficient to underwrite retention or concentration precisely. | 中 | SU003, SU004, SU023 |
| CU026 | Arm and AMD signals are strategically important because they reduce the odds that Eliyan is selling into an isolated technical niche. | 中 | SU005, SU006, SU026 |
| CU027 | Optical and networking participants such as Lumentum and Cisco imply Eliyan’s relevance may extend beyond intra-package use cases. | 中 | SU004, SU007, SU009 |
| CU028 | Customer-proof quality would improve materially if management can distinguish investor validation from actual purchase orders and shipped programs. | 中 | SU003, SU019 |
| CU029 | There is no public evidence of churn, complaints, or failed deployments in the retrieved set, but absence of evidence is not evidence of strong retention. | 中 | SU021, SU022 |
| CU030 | The product’s infrastructure depth means even one design win could be strategically large, making account quality more important than raw account count. | 中 | SU001, SU024 |
| CU031 | Evaluating or deploying language in the Series C announcement implies a funnel with at least two maturity levels, but no denominator is provided. | 中 | SU004 |
| CU032 | Third-party coverage suggests Eliyan is engaging both compute-side and optics-side constituencies, which supports a broader expansion narrative. | 中 | SU013, SU017 |
| CU033 | The strongest public customer positives are ecosystem relevance and strategic validation; the strongest negatives are retention opacity and concentration uncertainty. | 中 | SU019, SU022, SU023 |
| CU034 | Public customer evidence is therefore directionally positive but still early for a hardware company already marked at unicorn valuation. | 中 | SU015, SU021, SU024 |
| CU035 | Any investor should treat customer durability as a first-order diligence item before leaning heavily on the 2026 funding narrative. | 中 | SU020, SU023 |
| CR001 | Eliyan’s highest visible regulatory risk is export-control spillover because AI semiconductors are explicitly in scope of U.S. policy tightening. | 中 | SR006, SR007 |
| CR002 | Any expansion into optical and broader AI infrastructure does not remove exposure to semiconductor export-control regimes. | 中 | SR001, SR006 |
| CR003 | The public record does not disclose how Eliyan classifies products or manages export-control compliance. | 中 | SR001, SR006 |
| CR004 | No public litigation or enforcement history was surfaced in the retrieved set, but the legal-disclosure surface is thin because Eliyan is private. | 中 | SR008, SR030 |
| CR005 | Private-company opacity makes absence of public legal incidents weaker evidence than it would be for a public issuer. | 中 | SR008, SR009 |
| CR006 | Operational risk is elevated because Eliyan’s products sit inside complex chiplet, packaging, memory, and optics workflows. | 中 | SR005, SR017, SR018 |
| CR007 | Qualification and interoperability delays could materially slow revenue conversion even if technical benchmarks are attractive. | 中 | SR004, SR017, SR029 |
| CR008 | The public record does not disclose yield, defect, or field-reliability metrics. | 中 | SR004, SR005 |
| CR009 | Optical expansion adds dependency on counterparties and infrastructure layers that Eliyan does not fully control. | 中 | SR001, SR022, SR025 |
| CR010 | Arm, AMD, Cisco, and optics-related backers are assets, but they also highlight dependence on ecosystem relationships for commercialization. | 中 | SR021, SR023, SR024 |
| CR011 | Customer concentration risk is likely high because deep-infrastructure suppliers usually win a small number of large programs first. | 中 | SR004, SR017 |
| CR012 | That likely concentration can transmit directly into revenue lumpiness and valuation volatility. | 中 | SR016, SR019 |
| CR013 | Competition and consolidation risk are material because larger incumbents can bundle connectivity into broader platform offerings. | 中 | SR016, SR019 |
| CR014 | Standards momentum is a double-edged sword: it validates Eliyan’s category while reducing the defensibility of protocol support alone. | 中 | SR005, SR029 |
| CR015 | Financial/model risk remains elevated because public sources do not disclose revenue, gross margin, burn, or runway. | 中 | SR001, SR019 |
| CR016 | A $1B valuation increases the risk of future down-round pressure if design-win conversion lags market expectations. | 中 | SR019, SR020, SR026 |
| CR017 | Because fundraising remains central to the public story, Eliyan still appears dependent on external capital for scaled commercialization. | 中 | SR001, SR002 |
| CR018 | Key-person risk is meaningful because the company’s public narrative is tightly tied to founder credibility and technical leadership. | 中 | SR003, SR020 |
| CR019 | Hiring demand indicates active build-out, which is positive, but it also implies execution risk if recruiting or onboarding falls behind roadmap needs. | 中 | SR003 |
| CR020 | Process portability, standards work, and ecosystem ties mitigate some product risk, but none eliminates the need for production proof. | 中 | SR005, SR021, SR029 |
| CR021 | Broken or sparse public pages around some ecosystem references are themselves a diligence signal that third-party corroboration is uneven. | 低 | SR007, SR027, SR028 |
| CR022 | Comparable public companies disclose richer risk surfaces through filings, highlighting how much remains hidden in Eliyan’s private-company profile. | 中 | SR009, SR010, SR011, SR012, SR013, SR014, SR015 |
| CR023 | Supply-chain and manufacturing risk are likely important, but the public record does not identify specific foundries, OSAT partners, or single-source components in sufficient detail. | 中 | SR005, SR018 |
| CR024 | Any slip in electro-optical commercialization could weaken the strategic rationale supporting the 2026 financing. | 中 | SR001, SR019, SR026 |
| CR025 | The company’s strongest mitigants are capital raised, ecosystem validation, and technical specificity. | 中 | SR001, SR002, SR021 |
| CR026 | The company’s weakest public areas are legal/regulatory detail, operational metrics, and customer durability evidence. | 中 | SR003, SR006, SR019 |
| CR027 | No public evidence retrieved confirms formal quality certifications or structured incident disclosure. | 中 | SR004, SR005 |
| CR028 | The shift toward more distributed AI systems can expand demand while simultaneously increasing system-integration failure modes. | 中 | SR017, SR018 |
| CR029 | The company could face IP or standards disputes in the future simply because interconnect layers are strategically valuable and crowded, even though no active dispute surfaced in the retrieved set. | 低 | SR016, SR029 |
| CR030 | Customer proof remains insufficient to rule out long pilot cycles or stalled design evaluations. | 中 | SR001, SR019 |
| CR031 | If export-control scope expands further, customer mix and geographic go-to-market options could narrow. | 中 | SR006, SR007 |
| CR032 | If the category consolidates faster than Eliyan scales, the company may be forced into partnership, sale, or more difficult independent financing. | 中 | SR016, SR019 |
| CR033 | The absence of public debt or covenant data prevents a full balance-sheet risk assessment. | 中 | SR001, SR008 |
| CR034 | The combination of customer concentration risk and valuation pressure creates a clear thesis-break path if one or two flagship programs slip. | 中 | SR012, SR016, SR019 |
| CR035 | Regulatory diligence should focus first on export classification, sanctioned geographies, and any optical/module export dependencies. | 中 | SR006, SR022, SR025 |
| CR036 | Operational diligence should focus on qualification cycle times, defect rates, and foundry/packaging dependencies. | 中 | SR005, SR017, SR018 |
| CR037 | Customer diligence should focus on account concentration, evaluation-to-deployment conversion, and repeat orders. | 中 | SR004, SR019 |
| CR038 | Financing diligence should focus on cash runway relative to optical roadmap milestones and market adoption. | 中 | SR001, SR020 |
| CR039 | Despite many risks, there is no single surfaced showstopper; the issue is stacked uncertainty across several critical dimensions. | 中 | SR001, SR006, SR019 |
| CR040 | The investment case is therefore more exposed to execution and evidence gaps than to currently visible legal or compliance failures. | 中 | SR004, SR008, SR026 |
| CR041 | Even for comparable public companies, raw filing retrieval can be cumbersome, reinforcing that Eliyan-specific private diligence will need direct company cooperation rather than passive web review. | 低 | SR031 |
| CV001 | Eliyan’s latest priced reference point is a $145M Series C at a $1B valuation in July 2026. | 高 | SV001, SV002, SV004 |
| CV002 | That round followed a $50M strategic financing earlier in 2026, indicating rapid capital stacking rather than a long post-Series-B maturation period. | 中 | SV015, SV026 |
| CV003 | Public evidence does not disclose revenue, gross margin, burn, or backlog, so valuation cannot be triangulated to operating metrics with precision. | 中 | SV001, SV002 |
| CV004 | The current valuation therefore rests primarily on technical relevance, ecosystem validation, and expected future design-win conversion. | 中 | SV001, SV012, SV016, SV017 |
| CV005 | Independent market lenses support a structurally attractive category, but they do not by themselves justify Eliyan’s company-specific price. | 中 | SV008, SV009, SV010 |
| CV006 | Eliyan’s strongest valuation supports are the memory-wall problem, chiplet ecosystem growth, and expansion into electro-optical interconnects. | 中 | SV007, SV011, SV018 |
| CV007 | Its weakest valuation supports are missing operating metrics, customer-proof opacity, and commercialization timing risk. | 中 | SV002, SV013, SV014 |
| CV008 | A unicorn price without public revenue proof makes entry discipline essential. | 中 | SV001, SV003 |
| CV009 | The public record is strong enough to support a differentiated-company view, but not a clear buy-at-any-price view. | 中 | SV012, SV013 |
| CV010 | A reasonable base-case framing is that Eliyan deserves a premium to undifferentiated hardware startups, but a discount to fully proven infrastructure franchises. | 中 | SV002, SV008, SV013 |
| CV011 | Bull-case upside depends on converting technical ambition into repeated hyperscaler, accelerator, memory, and optical-adjacent wins before the next financing event. | 中 | SV001, SV015, SV016 |
| CV012 | Bear-case downside comes from slow qualification, concentration, and a future financing round that arrives before revenue proof. | 中 | SV013, SV014, SV019 |
| CV013 | The current evidence does not justify a strong-buy call because price sensitivity is high and operating disclosure is weak. | 中 | SV002, SV003, SV013 |
| CV014 | The valuation stance is best described as stretched rather than obviously impossible. | 中 | SV001, SV002, SV014 |
| CV015 | Recommendation quality is limited more by evidence gaps than by lack of market ambition. | 中 | SV003, SV012, SV013 |
| CV016 | Public-company filings from comparable infrastructure names remind investors how much disclosure would normally exist in a mature underwriting case. | 中 | SV019, SV023, SV024, SV025, SV035, SV036 |
| CV017 | Because Eliyan is private, the prudent recommendation should lean toward track or research-more unless management provides deeper financial and customer proof. | 中 | SV001, SV016, SV017 |
| CV018 | A $1B valuation may still be reasonable in a bull scenario if Eliyan becomes a critical layer in AI connectivity across multiple system domains. | 中 | SV006, SV007, SV018 |
| CV019 | That same valuation looks vulnerable in a bear scenario if the product remains strategically admired but commercially under-disclosed. | 中 | SV002, SV013, SV014 |
| CV020 | Ecosystem validation from AMD and Arm improves confidence that Eliyan is relevant to real design chains. | 中 | SV016, SV017 |
| CV021 | However, ecosystem validation is weaker than revenue or retention disclosure when setting price discipline. | 中 | SV016, SV017, SV029 |
| CV022 | The best-supported downside trigger is a financing event that arrives before management can produce repeat customer proof. | 中 | SV001, SV013, SV015 |
| CV023 | The best-supported upside trigger is public evidence that evaluations are turning into multiple production deployments. | 中 | SV001, SV007 |
| CV024 | Comparable valuation work remains weak because retrieved filing pages do not yield normalized revenue multiples from the public web surface used here. | 中 | SV019, SV023, SV024, SV025, SV035, SV036 |
| CV025 | That limitation pushes the analysis toward scenario valuation rather than hard multiple-based underwriting. | 中 | SV024, SV025, SV029 |
| CV026 | The current recommendation is more likely to improve after diligence than to worsen from category collapse alone. | 中 | SV008, SV009, SV018 |
| CV027 | Still, consolidation and bundling risk can compress the standalone value capture available to Eliyan. | 中 | SV013, SV014 |
| CV028 | A base-case investor should require evidence on customer conversion, concentration, and gross margin before underwriting appreciation from the current mark. | 中 | SV001, SV002, SV015 |
| CV029 | The round itself is a positive signal because sophisticated investors were willing to price the company at unicorn status in 2026, not in a weaker funding market. | 中 | SV002, SV004, SV027 |
| CV030 | But the same round also limits margin of safety for new investors if diligence uncovers slower-than-implied commercialization. | 中 | SV001, SV013 |
| CV031 | The most defendable near-term call is research-more rather than buy because the valuation is visible while the operating denominator is not. | 中 | SV002, SV003, SV024 |
| CV032 | Confidence should remain medium rather than high because the case relies on inferred future conversion instead of disclosed present economics. | 中 | SV002, SV003, SV016 |
| CV033 | Risk rating is high because concentration, qualification, export-control, and evidence gaps can all interact. | 中 | SV013, SV014, SV029 |
| CV034 | If the company later shows repeat deployments and healthy gross margins, the same $1B mark could look conservative rather than stretched. | 中 | SV007, SV018 |
| CV035 | If customer proof remains indirect through the next financing cycle, the current mark will likely look expensive. | 中 | SV013, SV015 |
| CV036 | Public evidence supports a non-zero probability of strategic exit interest because the company sits in a strategically important layer for larger infrastructure players. | 中 | SV006, SV013, SV018 |
| CV037 | Exit readiness is constrained today by limited public proof of durable revenue and customer scale. | 中 | SV002, SV024 |
| CV038 | The underwriting method most consistent with the evidence is milestone-sensitive scenario analysis, not conventional multiple comparison. | 中 | SV024, SV025 |
| CV039 | Therefore the final call should be price-sensitive, evidence-sensitive, and explicit about what would move it to buy or avoid. | 中 | SV028, SV029 |
| CV040 | On current public evidence alone, Eliyan looks like a company to track and diligence hard rather than to underwrite aggressively at the visible mark. | 中 | SV001, SV013 |