近日，中国石油天然气集团有限公司2026年网络设备（信创产品第二批次）带量集中采购正式挂网招标。作为年内能源领域规模领先的信创设备集采项目，本次共设17个标包，涵盖台式机、终端、服务器三大品类，总采购量约1.3万台/套，预算规模超4亿元。

Recently, the China National Petroleum Corporation officially launched the online bidding for its 2026 large-scale centralized procurement of network equipment (the second batch of information technology innovation products). As the largest information technology innovation equipment centralized procurement project in the energy sector this year, this round of bidding has set up 17 packages, covering three categories: desktop computers, terminals, and servers. The total procurement volume is approximately 13,000 units/sets, with a budget exceeding 400 million yuan.

当前，能源行业国产化进程从办公系统替代走向生产系统替代，本次集采的技术路线选择，也成为观察国产算力产业工业级落地能力的重要窗口。随着集采推进，海光信息基于C86架构的整机方案在多个标包中入围，实现全品类覆盖，对应采购金额占比超53%，是本次集采中布局最完整的技术路线。

Currently, the domesticization process in the energy industry is shifting from replacing office systems to replacing production systems. The technical route selection for this centralized procurement has become an important window for observing the industrial-level implementation capabilities of the domestic computing power industry. As the centralized procurement progresses, Hai Guang Information's complete machine solution based on the C86 architecture has been selected in multiple bidding packages, achieving full coverage of all categories, with the corresponding procurement amount accounting for over 53%, making it the most comprehensive technical route in this centralized procurement.

全品类统一架构落地，C86路线破解规模化国产化痛点

The unified architecture for all product categories has been implemented, and the C86 route has solved the problems of large-scale domestic production.

此前，国内信创采购多集中在桌面终端与办公软件等通用场景，普遍存在技术路线分散、生态体系割裂的问题。随着央国企国产化进入深水区，采购重心逐步向核心算力、生产业务系统延伸，对厂商的全栈产品能力与生态适配能力提出了更高要求。

Previously, domestic information technology innovation procurement mainly focused on desktop terminals and office software in common scenarios, and there were widespread problems such as fragmented technical routes and disconnected ecosystems. As the domesticization of central and state-owned enterprises entered the deep-water zone, the procurement focus gradually extended to core computing power and production business systems, which placed higher demands on the full-stack product capabilities and ecosystem adaptation capabilities of manufacturers.

从本次集采的标包划分来看，国内多条主流算力技术路线同台竞技，海光C86架构是其中覆盖品类最完整、适配场景最丰富的技术路线之一。

From the division of the bidding packages in this round of centralized procurement, multiple mainstream computing power technology routes from China are competing on the same stage. The Haidong C86 architecture is one of the technology routes that covers the most complete categories and has the richest application scenarios.

工作站品类下，基于海光芯片的工作站是本次集采唯一入选的技术路线，相关方案支持国内油气专业软件以及国际主流软件，满足油气勘探数据处理、地质建模、材料仿真等专业场景的高强度计算需求；桌面侧，海光芯片台式机基于对x86生态的兼容性，可平滑对接中石油存量业务系统与工业软件，大幅降低业务迁移成本与适配门槛；服务器端，C86架构通算服务器采购规模在标包总预算中占比过半，能够同时承载综合办公、经营管理与生产运营类业务，实现统一架构下的多负载复用。

Among the workstation categories, the workstation based on Haidong chips is the only technical route selected in this centralized procurement. The related solutions support domestic oil and gas professional software as well as international mainstream software, meeting the high-intensity computing requirements of professional scenarios such as oil and gas exploration data processing, geological modeling, and material simulation. On the desktop side, the Haidong chip desktop computers, based on compatibility with the x86 ecosystem, can smoothly connect with the existing business systems and industrial software of PetroChina, significantly reducing the cost of business migration and the threshold of adaptation. On the server side, the C86 architecture all-inclusive server procurement scale accounts for more than half of the total budget of the package, capable of simultaneously supporting comprehensive office, management and production operation businesses, and achieving multi-load reuse under a unified architecture.

业内分析师指出，能源央企业务链条长、场景差异大，统一架构的全品类覆盖，意味着企业可以构建从前端桌面到后端数据中心的一致算力底座，大幅减少跨架构适配的研发投入与运维成本。这也是头部央企在信创进入深水区后，更倾向于选择具备全栈能力厂商的核心逻辑。

Industry analysts point out that energy central enterprises have long business chains and diverse scenarios. A unified architecture covering all categories means that the enterprises can build a consistent computing infrastructure from the front-end desktop to the back-end data center, significantly reducing the R&D investment and operation and maintenance costs for cross-architecture adaptation. This is also the core logic that leading central enterprises prefer when choosing core vendors with full-stack capabilities after the informatization and innovation initiative enters the deep-water zone.

双芯算力渗透核心生产，能源国产化进入价值验证阶段

Dual-core computing power penetrates the core production process, and the domestication of energy resources has entered the stage of value verification.

值得关注的是，本次集采不再局限于通用办公场景，而是将智算与高端计算同步纳入采购范围，采购需求已从“桌面替代”延伸到产业链核心业务环节。

It is worth noting that this round of centralized procurement is no longer limited to general office scenarios. Instead, intelligent computing and high-end computing have been included in the procurement scope simultaneously. The procurement requirements have expanded from "desktop replacement" to the core business links of the entire industrial chain.

通算服务器的规模化落地应用，是验证国产算力能否深度支撑核心生产系统的关键标尺，其核心价值在于以统一架构同时承载综合办公、经营管理、生产运营和高端计算等多元业务负载。

The large-scale application of the computing server is a key indicator for verifying whether the domestic computing power can deeply support core production systems. Its core value lies in being able to simultaneously handle various business loads such as comprehensive office work, management and operation, production and operation, and high-end computing through a unified architecture.

实际应用层面，海光通算服务器已在中石油多家二级单位完成落地验证。依托C86架构的通用算力与生态兼容能力，油田勘探开发中的地震资料处理、油藏数值模拟、炼化场景中的流程模拟与工艺优化、销售系统中的加油站管理，多种高端计算任务均可在同一算力底座上运行，减少跨架构调度和数据孤岛。

In practical application, the Haiguang computing server has been successfully implemented and verified in multiple secondary units of China National Petroleum Corporation. Leveraging the general computing power and ecosystem compatibility of the C86 architecture, various high-end computing tasks such as seismic data processing in oilfield exploration and development, oil reservoir numerical simulation, process simulation and process optimization in the refining and chemical industry, and gas station management in the sales system can all be executed on the same computing power platform, reducing cross-architecture scheduling and data silos.

智算层面，在中石油此前的智算服务器采购中，海光DCU已完成95余款主流大模型的适配验证，覆盖语言、视觉、多模态、科学计算等全类别，支撑昆仑大模型在集团内部多场景落地。本次集采延续了这一技术路径，C86智算服务器可同时覆盖AI大模型训练推理与工业科学计算需求，适配核心科研生产场景。

At the intelligent computing level, in the previous intelligent computing server procurement of China National Petroleum Corporation, Haiguang DCU has completed the adaptation and verification of over 95 mainstream large models, covering all categories such as language, vision, multimodal, and scientific computing, supporting the Kunlun large model to be implemented in multiple scenarios within the group. This centralized procurement continues this technical path. The C86 intelligent computing server can simultaneously cover the training and inference of AI large models as well as industrial scientific computing needs, and is compatible with core scientific research and production scenarios.

作为能源领域的标杆性集采项目，中石油本次的技术路线选择具备较强的行业示范效应，也印证了国产算力已具备支撑关键行业核心生产业务的能力。随着能源行业数字化与智能化转型持续深化，核心生产场景的算力国产化需求将持续释放，推动能源信创从“可用”向“好用、管用”加速演进。

As a benchmark centralized procurement project in the energy sector, the technical route chosen by PetroChina this time has a strong industry demonstration effect and also confirms that domestic computing power has the ability to support the core production business of key industries. With the continuous deepening of the digital and intelligent transformation in the energy industry, the demand for the domesticization of computing power in core production scenarios will continue to be released, promoting the energy information innovation and transformation to accelerate from "availability" to "usability and manageability".