A post on X by @Priyannkaaaa reports that DeepSeek has “closed in on” more than $12 billion from Tencent and CATL, alongside plans for a large Huawei-based computing expansion for a possible V5 model. It also says DeepSeek V5 could arrive “next month,” but the post labels that timing unconfirmed and gives no date anchor for identifying which month it means.
The funding figure, infrastructure plans and possible launch should be read as details reported by the post, not as confirmed announcements. The post does not establish that a financing round was completed, that the proposed hardware has been deployed or that DeepSeek has scheduled a V5 release.
What the DeepSeek V5 report says
According to the X post, DeepSeek has “closed in on” more than $12 billion from Tencent and CATL, surpassing an original $7.5 billion target. It says the capital is intended for compute expansion, model development and a possible 2027 initial public offering, or IPO—the process of offering shares in a company to public investors.
That wording does not establish that DeepSeek completed a financing round or that the original target formally existed. The post also does not establish that a 2027 IPO plan has been approved or filed.
The post says DeepSeek V5 would be the company’s first model trained primarily on domestic Huawei silicon instead of Nvidia hardware. It provides no model specifications, benchmark results, release materials or evidence that training on this hardware has begun.
Image credit: @Priyannkaaaa on X
The graphic supports only the visible “deepseek v5” label. It does not verify the funding, chip deployment, training plan or launch timing described in the post.
Reported funding and intended uses
The post links the reported capital to three intended uses:
expanding computing capacity;
developing models; and
supporting a possible 2027 IPO.
The post does not specify whether these are committed budgets, proposed uses of capital or a mixture of both. Until an involved company or an official filing confirms the amount and its purpose, the more-than-$12-billion figure remains a reported financing detail rather than a verified measure of DeepSeek’s resources.
The proposed Huawei Ascend computing plan
The post says DeepSeek plans to deploy more than 160,000 Huawei Ascend accelerators at a new data center in Inner Mongolia. It further describes V5 as being trained primarily on Huawei hardware rather than Nvidia hardware.
Accelerators are specialized chips used to perform the large numbers of calculations required by AI training and inference. Training adjusts a model using data, while inference is the later process of running the trained model to produce outputs. A large deployment could support either stage or both, but the report does not explain the facility’s technical design, delivery schedule or operating status.
The Inner Mongolia location, the number of accelerators and the proposed Huawei-based training plan therefore remain unconfirmed details from the post. They should not be treated as evidence that the hardware has already been installed or that DeepSeek V5 has been trained on it.
What the reported chip allocation could indicate
The post describes a roughly 80/20 split between training and inference. If accurate, that allocation would indicate that most of the proposed accelerator capacity is intended for developing or updating models, with a smaller share reserved for running models for users or internal workloads.
The ratio is a description of a reported plan, not a measured operating result. The post does not explain how the split was calculated, whether it applies to the entire proposed deployment or whether it could change after launch. It also supplies no information about model size, training duration, user demand or actual performance.
Is DeepSeek V5 expected to launch soon?
The post says DeepSeek V5 could arrive “next month,” but labels that timing unconfirmed. Because the post provides no date anchor, readers cannot determine which month the phrase refers to. It also gives no exact date, official release announcement, availability information or evidence that the model is ready for public use.
The possible launch is therefore a reported possibility, not a release schedule. The same distinction applies to the proposed IPO and infrastructure plans: the post describes what may happen, but those plans are not confirmed announcements.
How to read the report until official confirmation
The clearest status-based reading is:
Reported by the post: DeepSeek has “closed in on” more than $12 billion from Tencent and CATL; it may build a data center in Inner Mongolia; it may deploy more than 160,000 Huawei Ascend accelerators; and the proposed allocation is roughly 80% for training and 20% for inference.
Described as plans: compute expansion, model development, primarily Huawei-based training and a possible 2027 IPO.
Explicitly unconfirmed: a V5 launch “next month.” The post provides no date anchor, so the timing cannot be tied to a specific month from the supplied information.
Not established by the report: a completed financing round, installed accelerators, completed Huawei-based training, model specifications, performance results or a confirmed availability date.
Until DeepSeek or the other named organisations provide supporting information, the report is best understood as an unverified account of possible funding, infrastructure and launch plans—not confirmation that the funding has closed, the chips have been deployed or DeepSeek V5 has a scheduled release.
0 comments
No approved comments yet. You can start the conversation.
Leave a comment