供资
大气AI 安全 10M种子用于采购模型推
在 Google 上的首选来源中添加 Unite.AI
大气AI 宣布在2026年11月5日表示，它拥有由DBL Partners牵头的1000万美元的融资种子资金，以扩大其原材料供应链的计算模型。 DBL Partners于2026年11月4日公布了其组合页面。 DBL的Vinnie Caruso 加入的棋盘也开始移动。
这笔资金还将深化与首批大约100个制造商的合作。 这一扩展将技术工作与商业推动联系起来：托尔斯滕·利希特瑙博士在拜仁公司工作了22年之后，作为创始首席商业官员加入其中，他是该公司的高级合作伙伴和全球碳转化负责人。
示范工业供应链
大气AI的方法始于制作材料所使用的物理过程，而不是依赖供应商披露内部操作数据。 其模型包括化学工程、植物经济学和工业工艺，包括原料、能源、生产成本和排放之间的关系。 连接系统包括油田和管道、炼油厂、化工厂、码头和航运路线。
大气AI 称运行着约5万个设施模型，在一次测试中，将一个生产者公布的碳足迹复制到了4.4%以内，而没有使用该生产者自己的数据。 这些模型旨在估计材料的成本，评估供应商的价格上涨是否合理，确定其他可提供材料的工厂，并审查中断的路线如何影响成本、碳或复原力。
公司的产品描述将平台呈现为从原产地到工厂大门的原材料价值链的动态数字双胞胎。 具体设施成本、碳和风险信息支持对单个供应商进行成本分析、针对特定供应商进行范围3排放计量和生命周期评估一体化。 多层次的绘图、主动预警和来源建议旨在将干扰和谈判决定纳入同一工作流程。
对采购决定适用模式
Haleon是一家已经与大气AI合作的制造商。 称公司在树脂采购方面合作了近两年。 他的团队现在从材料的应成本、碳足迹和供应链风险开始，而不是仅仅依赖供应商的报价，将模型制作工作与购买决定直接联系起来。
在大气AI介绍的一个客户例子中，一家制造商在Hormuz干扰期间面临每吨730美元的拟议价格上涨。 启动小组计算，制造成本增加大约为每吨120美元。 这一分析为采购小组提供了一个基础，讨论生产经济学的变化，而不是简单地接受提案或无证据地提出异议。
共同创始人兼CEO罗伊·戈德伯格(Roee Goldberg)将长期目标描述为原材料经济的计算模型：制造商会了解他们所购买的工厂和供应网络，化学工程提供了基础，AI为作品提供了更大规模的能力。
该回合规定的直接检验标准是宽度：已经用于树脂采购的基于设施的办法是否能够支持跨其他材料类别的采购决定。 大气AI正在将这种扩展与新的商业领导和更深层次的制造商合作结合起来，使超越塑料的移动成为种子资金所要达到的目的。
本文由 AI图鉴自动抓取并翻译，版权归原作者所有；如需查看原文请点下方链接。
Funding
Atmospheric AI Secures $10M Seed for Procurement Modeling Push
Add Unite.AI to your preferred sources on Google
Atmospheric AI announced on November 5, 2026, that it had raised $10 million in seed funding led by DBL Partners to expand its computational models of raw-material supply chains. DBL Partners published its portfolio page on November 4, 2026. The startup plans to move beyond plastics into chemicals, paper and metals, while DBL’s Vinnie Caruso joins its board.
The funding will also deepen collaboration with an initial group of around 100 manufacturers. That expansion connects the technical work to a commercial push: Dr Torsten Lichtenau is joining as founding chief commercial officer after 22 years at Bain & Company, where he was a senior partner and global head of carbon transition.
Modeling Industrial Supply Chains
Atmospheric AI’s approach begins with the physical processes used to make materials, rather than depending on suppliers to disclose internal operating data. Its models incorporate chemical engineering, plant economics and industrial processes, including relationships among feedstocks, energy, production costs and emissions. The connected system covers oil fields and pipelines, refineries, chemical plants, terminals and shipping routes.
Atmospheric AI says it runs around 50,000 facility models and, in one test, reproduced a producer’s published carbon footprint to within 4.4% without using that producer’s own data. The models are intended to estimate what a material should cost, assess whether a supplier’s price increase is justified, identify other plants that could supply it and examine how a disrupted route could affect cost, carbon or resilience.
The company’s product description presents the platform as a dynamic digital twin of a raw-material value chain, from its origin to the factory gate. Facility-specific cost, carbon and risk information supports should-cost analysis for individual suppliers, supplier-specific Scope 3 emissions metrics and lifecycle-assessment integration. Multi-tier mapping, proactive alerts and sourcing recommendations are intended to bring disruption and negotiation decisions into the same workflow.
Applying Models to Procurement Decisions
Haleon is one manufacturer already working with Atmospheric AI. Chris Low, Haleon’s head of sustainability procurement, said the companies have collaborated for almost two years on resin sourcing. His team now begins with a material’s should-cost, carbon footprint and supply-chain risks rather than relying only on the supplier’s quote, connecting the modeling work directly to a buying decision.
In a customer example described by Atmospheric AI, a manufacturer faced a proposed price increase of $730 per tonne during the Hormuz disruption. The startup’s team calculated that the underlying increase in manufacturing costs was approximately $120 per tonne. That analysis gave the sourcing team a basis for discussing what had changed in production economics instead of simply accepting the proposal or disputing it without evidence.
Co-founder and CEO Roee Goldberg described the long-term goal as a computational model of the raw-material economy: manufacturers would understand the plants and supply networks behind what they buy, with chemical engineering providing the foundation and AI enabling the work at a much larger scale.
The immediate test set by the round is breadth: whether a facility-based approach already applied to resin sourcing can support procurement decisions across additional material categories. Atmospheric AI is pairing that expansion with new commercial leadership and deeper manufacturer collaboration, making the move beyond plastics central to what the seed funding is meant to enable.
Auto-fetched and translated by AI Atlas. All rights belong to the original author.