COMPANIES · FOUNDATION MODELS

DeepSeek

DeepSeek is a Chinese AI lab, spun out of the quantitative hedge fund High-Flyer, that develops and open-sources the DeepSeek-V and DeepSeek-R families of large language models.

DirectFoundation models
What it does

DeepSeek is a Chinese AI lab, spun out of the quantitative hedge fund High-Flyer, that develops and open-sources the DeepSeek-V and DeepSeek-R families of large language models.

Why it matters

DeepSeek's January 2025 release of its R1 model, reported to have been trained at unusually low cost, prompted a broad reassessment of AI compute spending and a sharp, short-term drop in NVIDIA's stock price.

How it makes money

DeepSeek offers low-priced API access to its models and open-sources model weights; it is privately funded by High-Flyer and discloses no independent revenue figures.

Geography

China (Hangzhou)

AI stack position

Foundation models and AI laboratories. The organizations actually building the AI models themselves, like the systems behind ChatGPT, Claude or Gemini. This is the most crowded, fastest-moving layer of the AI economy right now.

Connections (5)

Every connection below is tied to a readable source, date, classification and statement of its limits. A company source establishes what that company stated; it does not necessarily provide independent confirmation.

Current relationships (1)

DeepSeek publishes models through Amazon Web Services
Publishes throughCurrent
DeepSeek-V3.1, a 685-billion-parameter hybrid reasoning model, was made available globally as a fully managed model in Amazon Bedrock.

Source: 2025-09-18 · reviewed 2026-09-22

Evidence: Primary-source statement · Reported operating

Limits: Does not establish any formal agreement or payment between DeepSeek and AWS beyond AWS hosting the open-weight model; DeepSeek has not itself confirmed involvement.

Alibaba's Qwen3 and DeepSeek-V3.1 models now available in Amazon Bedrock ↗

Dated or announced (1)

DeepSeek trained early models on NVIDIA GPUs NVIDIA
Uses technology fromDated / announced
Reporting states DeepSeek's earlier models were trained using NVIDIA GPUs (including H800 chips) before the company began shifting significant training and inference work toward domestic Huawei Ascend hardware in 2026.

Source: 2026-07-07 · reviewed 2026-09-20

Evidence: Secondary reporting (unconfirmed by primary parties) · Reported shifting away

Limits: Does not establish DeepSeek's current primary hardware supplier, purchase volumes, or whether any NVIDIA relationship is ongoing as of the review date.

Exclusive-China's DeepSeek Developing Its Own AI Chip, Sources Say ↗

Comparisons (3)

Tencent competes with DeepSeek
Competes withComparison only
Tencent publicly released a Hunyuan Turbo model that it stated performed faster than DeepSeek's model on certain measures, per Bloomberg reporting of the on-record claim.

Source: 2025-02-27 · reviewed 2026-09-20

Evidence: Secondary reporting of an on-record executive statement · Reported operating

Limits: Reflects Tencent's own comparative claim about a specific model release, not an independent benchmark or a statement about overall competitive position.

Tencent Fires Up AI Race With Model It Says Outdoes DeepSeek ↗
DeepSeek competes with OpenAI
Competes withComparison only
DeepSeek's low-cost, open-weight models are widely reported as directly competing with OpenAI's and other US labs' proprietary models on cost and capability.

Source: Source undated · reviewed 2026-09-20

Evidence: Analytical comparison · Reported operating

Limits: Based on overlapping product categories (general-purpose LLMs), not on a market-share or revenue comparison.

DeepSeek | Rise, Technologies, Impact, & Global Response ↗
Moonshot AI competes with DeepSeek
Competes withComparison only
Industry coverage groups Moonshot AI's Kimi models with DeepSeek's models as competing entrants in China's open-weight large language model landscape.

Source: Source undated · reviewed 2026-09-20

Evidence: Analytical comparison · Reported operating

Limits: Reflects overlapping product categories rather than a claim from either company about the other.

China's Top AI Models in 2026: DeepSeek, Qwen, Kimi, Doubao and the New AI Race ↗
Evidence for this profile

DeepSeek | Rise, Technologies, Impact, & Global Response ↗
Reviewed

Explore it interactively

Open this profile in the database See it on the relationship map

Inclusion in this database is not an endorsement of a company or its securities. Nothing here is a recommendation to buy or sell anything, and none of it is individualized investment advice. See the disclosures. Data: AI Map v3.22, released .