DeepSeek R1's Implications: Winners and Losers in the Generative AI Value Chain

R1 is mainly open, on par with leading proprietary models, appears to have been trained at considerably lower expense, and is cheaper to use in terms of API gain access to, all of which point to an innovation that may change competitive dynamics in the field of Generative AI.
- IoT Analytics sees end users and AI applications companies as the most significant winners of these recent developments, while exclusive model suppliers stand to lose the most, based upon worth chain analysis from the Generative AI Market Report 2025-2030 (released January 2025).

Why it matters

For suppliers to the generative AI worth chain: Players along the (generative) AI worth chain might need to re-assess their value proposals and line up to a possible truth of low-cost, lightweight, open-weight models.
For generative AI adopters: DeepSeek R1 and other frontier models that might follow present lower-cost options for AI adoption.

Background: DeepSeek's R1 model rattles the marketplaces

DeepSeek's R1 model rocked the stock markets. On January 23, 2025, China-based AI start-up DeepSeek launched its open-source R1 reasoning generative AI (GenAI) model. News about R1 quickly spread, and by the start of stock trading on January 27, 2025, the market cap for numerous significant technology companies with large AI footprints had actually fallen considerably because then:

NVIDIA, a US-based chip designer and developer most known for its information center GPUs, dropped 18% between the marketplace close on January 24 and the marketplace close on February 3.
Microsoft, the leading hyperscaler in the cloud AI race with its Azure cloud services, dropped 7.5% (Jan 24-Feb 3).
Broadcom, a semiconductor business focusing on networking, broadband, and customized ASICs, dropped 11% (Jan 24-Feb 3).
Siemens Energy, a German energy innovation supplier that supplies energy services for information center operators, dropped 17.8% (Jan 24-Feb 3).

Market participants, and particularly financiers, responded to the narrative that the model that DeepSeek launched is on par with innovative designs, was allegedly trained on only a couple of countless GPUs, and is open source. However, because that preliminary sell-off, reports and analysis shed some light on the preliminary buzz.

The insights from this short article are based upon

Download a sample for more information about the report structure, select definitions, choose market information, additional information points, and patterns.

DeepSeek R1: What do we know until now?

DeepSeek R1 is a cost-effective, advanced reasoning design that measures up to top rivals while fostering openness through publicly available weights.

DeepSeek R1 is on par with leading reasoning designs. The biggest DeepSeek R1 model (with 685 billion parameters) efficiency is on par or even much better than a few of the leading models by US structure design suppliers. Benchmarks reveal that DeepSeek's R1 design performs on par or better than leading, more familiar designs like OpenAI's o1 and Anthropic's Claude 3.5 Sonnet.
DeepSeek was trained at a considerably lower cost-but not to the degree that initial news recommended. Initial reports suggested that the training costs were over $5.5 million, however the real worth of not only training however developing the model overall has actually been debated because its release. According to semiconductor research study and consulting firm SemiAnalysis, the $5.5 million figure is just one aspect of the expenses, leaving out hardware spending, the salaries of the research and development team, and other elements.
DeepSeek's API prices is over 90% more affordable than OpenAI's. No matter the true cost to establish the design, DeepSeek is providing a more affordable proposal for utilizing its API: input and output tokens for DeepSeek R1 cost $0.55 per million and $2.19 per million, respectively, compared to OpenAI's $15 per million and $60 per million for its o1 model.
DeepSeek R1 is an innovative model. The related scientific paper launched by DeepSeekshows the approaches used to develop R1 based on V3: leveraging the mix of specialists (MoE) architecture, reinforcement knowing, and very creative hardware optimization to create models requiring less resources to train and likewise less resources to carry out AI inference, causing its aforementioned API usage expenses.
DeepSeek is more open than the majority of its rivals. DeepSeek R1 is available free of charge on platforms like HuggingFace or GitHub. While DeepSeek has made its weights available and provided its training methods in its research study paper, the original training code and data have not been made available for an experienced person to develop a comparable model, aspects in defining an open-source AI system according to the Open Source Initiative (OSI). Though DeepSeek has actually been more open than other GenAI business, R1 remains in the open-weight classification when considering OSI requirements. However, the release triggered interest outdoors source community: Hugging Face has introduced an Open-R1 initiative on Github to develop a full reproduction of R1 by building the "missing pieces of the R1 pipeline," moving the model to fully open source so anybody can reproduce and build on top of it.
DeepSeek launched powerful little models together with the major R1 release. DeepSeek launched not only the major large model with more than 680 billion parameters however also-as of this article-6 distilled models of DeepSeek R1. The designs vary from 70B to 1.5 B, the latter fitting on lots of consumer-grade hardware. Since February 3, 2025, the designs were downloaded more than 1 million times on HuggingFace alone.
DeepSeek R1 was perhaps trained on OpenAI's data. On January 29, 2025, reports shared that Microsoft is examining whether DeepSeek used OpenAI's API to train its designs (an infraction of OpenAI's terms of service)- though the hyperscaler likewise added R1 to its Azure AI Foundry service.

Understanding the generative AI worth chain

GenAI spending advantages a broad industry value chain. The graphic above, based on research study for IoT Analytics' Generative AI Market Report 2025-2030 (released January 2025), portrays crucial beneficiaries of GenAI costs across the value chain. Companies along the value chain include:

Completion users - End users consist of customers and services that utilize a Generative AI application.
GenAI applications - Software suppliers that include GenAI features in their items or deal standalone GenAI software application. This consists of business software business like Salesforce, with its focus on Agentic AI, and startups specifically focusing on GenAI applications like Perplexity or Lovable.
Tier 1 beneficiaries - Providers of foundation designs (e.g., OpenAI or Anthropic), design management platforms (e.g., AWS Sagemaker, Google Vertex or Microsoft Azure AI), information management tools (e.g., MongoDB or Snowflake), cloud computing and data center operations (e.g., Azure, AWS, Equinix or Digital Realty), AI experts and combination services (e.g., Accenture or Capgemini), and edge computing (e.g., Advantech or HPE).
Tier 2 beneficiaries - Those whose product or services frequently support tier 1 services, consisting of providers of chips (e.g., NVIDIA or AMD), network and server devices (e.g., Arista Networks, Huawei or Belden), server cooling technologies (e.g., Vertiv or Schneider Electric).
Tier 3 beneficiaries - Those whose services and products frequently support tier 2 services, such as suppliers of electronic design automation software suppliers for chip design (e.g., Cadence or Synopsis), semiconductor fabrication (e.g., TSMC), heat exchangers for cooling technologies, and electrical grid technology (e.g., Siemens Energy or ABB).
Tier 4 recipients and beyond - Companies that continue to support the tier above them, such as lithography systems (tier-4) necessary for semiconductor fabrication machines (e.g., AMSL) or business that supply these providers (tier-5) with lithography optics (e.g., Zeiss).

Winners and losers along the generative AI worth chain

The rise of designs like DeepSeek R1 signals a prospective shift in the generative AI value chain, challenging existing market characteristics and reshaping expectations for success and competitive advantage. If more models with similar capabilities emerge, certain gamers might benefit while others deal with increasing pressure.

Below, IoT Analytics assesses the crucial winners and likely losers based upon the developments presented by DeepSeek R1 and the broader trend towards open, affordable designs. This assessment considers the possible long-lasting impact of such designs on the value chain rather than the instant results of R1 alone.

Clear winners

End users

Why these innovations are positive: The availability of more and cheaper designs will eventually reduce expenses for the end-users and make AI more available.
Why these innovations are unfavorable: No clear argument.
Our take: DeepSeek represents AI innovation that eventually benefits the end users of this technology.

GenAI application suppliers

Why these developments are favorable: Startups building applications on top of foundation designs will have more options to pick from as more designs come online. As mentioned above, DeepSeek R1 is by far cheaper than OpenAI's o1 design, and though reasoning models are hardly ever utilized in an application context, it reveals that ongoing advancements and innovation enhance the designs and make them more affordable.
Why these developments are negative: No clear argument.
Our take: The availability of more and more affordable designs will ultimately reduce the cost of consisting of GenAI functions in applications.

Likely winners

Edge AI/edge calculating business

Why these developments are positive: During Microsoft's recent profits call, Satya Nadella explained that "AI will be far more common," as more work will run locally. The distilled smaller sized models that DeepSeek released along with the powerful R1 design are small enough to operate on lots of edge gadgets. While small, the 1.5 B, 7B, and 14B models are also comparably effective reasoning designs. They can fit on a laptop computer and other less powerful devices, e.g., IPCs and industrial gateways. These distilled designs have actually already been downloaded from Hugging Face hundreds of countless times.
Why these innovations are negative: No clear argument.
Our take: The distilled models of DeepSeek R1 that fit on less powerful hardware (70B and below) were downloaded more than 1 million times on HuggingFace alone. This shows a strong interest in deploying designs locally. Edge computing producers with edge AI options like Italy-based Eurotech, and Taiwan-based Advantech will stand to profit. Chip business that concentrate on edge computing chips such as AMD, ARM, Qualcomm, and even Intel, might likewise benefit. Nvidia likewise runs in this market sector.

Note: IoT Analytics' SPS 2024 Event Report (released in January 2025) looks into the most current commercial edge AI trends, as seen at the SPS 2024 fair in Nuremberg, Germany.

Data management companies

Why these innovations are favorable: There is no AI without data. To establish applications using open models, adopters will require a huge selection of information for training and throughout implementation, needing correct data management.
Why these innovations are negative: No clear argument.
Our take: Data management is getting more crucial as the number of various AI designs increases. Data management companies like MongoDB, Databricks and Snowflake as well as the particular offerings from hyperscalers will stand to profit.

GenAI companies

Why these developments are favorable: The unexpected introduction of DeepSeek as a top player in the (western) AI environment shows that the intricacy of GenAI will likely grow for some time. The higher availability of different models can cause more complexity, driving more demand for services.
Why these developments are unfavorable: When leading designs like DeepSeek R1 are available for free, the ease of experimentation and application may limit the need for integration services.
Our take: As brand-new developments pertain to the market, GenAI services demand increases as business attempt to comprehend how to best utilize open designs for their company.

Neutral

Cloud computing providers

Why these developments are favorable: Cloud gamers hurried to include DeepSeek R1 in their model management platforms. Microsoft included it in their Azure AI Foundry, and AWS allowed it in Amazon Bedrock and Amazon Sagemaker. While the hyperscalers invest greatly in OpenAI and Anthropic (respectively), they are also model agnostic and make it possible for hundreds of various models to be hosted natively in their design zoos. Training and fine-tuning will continue to occur in the cloud. However, as models end up being more effective, less investment (capital investment) will be required, which will increase revenue margins for hyperscalers.
Why these innovations are negative: More designs are expected to be deployed at the edge as the edge ends up being more powerful and models more effective. Inference is likely to move towards the edge moving forward. The expense of training advanced models is likewise anticipated to decrease further.
Our take: Smaller, more effective models are ending up being more vital. This reduces the need for powerful cloud computing both for training and inference which might be offset by higher overall demand and lower CAPEX requirements.

EDA Software providers

Why these developments are favorable: Demand for new AI chip styles will increase as AI work end up being more specialized. EDA tools will be important for developing effective, smaller-scale chips tailored for edge and dispersed AI inference
Why these developments are unfavorable: The approach smaller sized, less resource-intensive models might minimize the need for developing advanced, high-complexity chips enhanced for enormous information centers, possibly leading to lowered licensing of EDA tools for high-performance GPUs and ASICs.
Our take: EDA software companies like Synopsys and Cadence could benefit in the long term as AI specialization grows and drives need for new chip styles for edge, consumer, and affordable AI workloads. However, the industry might require to adjust to moving requirements, focusing less on big information center GPUs and more on smaller, efficient AI hardware.

Likely losers

AI chip business

Why these developments are positive: The supposedly lower training costs for designs like DeepSeek R1 could ultimately increase the total need for AI chips. Some referred to the Jevson paradox, the concept that effectiveness causes more require for a resource. As the training and inference of AI designs become more efficient, the need might increase as higher effectiveness causes lower expenses. ASML CEO Christophe Fouquet shared a similar line of thinking: "A lower cost of AI might indicate more applications, more applications suggests more need in time. We see that as an opportunity for more chips need."
Why these developments are negative: The apparently lower expenses for DeepSeek R1 are based mainly on the need for less cutting-edge GPUs for training. That puts some doubt on the sustainability of large-scale tasks (such as the recently revealed Stargate project) and the capital investment costs of tech companies mainly earmarked for buying AI chips.
Our take: IoT Analytics research for its latest Generative AI Market Report 2025-2030 (published January 2025) discovered that NVIDIA is leading the information center GPU market with a market share of 92%. NVIDIA's monopoly defines that market. However, that likewise demonstrates how strongly NVIDA's faith is linked to the continuous development of spending on information center GPUs. If less hardware is needed to train and deploy designs, then this might seriously deteriorate NVIDIA's development story.

Other classifications related to information centers (Networking equipment, electrical grid innovations, electricity providers, and heat exchangers)

Like AI chips, designs are likely to become cheaper to train and more effective to release, so the expectation for more information center infrastructure build-out (e.g., networking equipment, cooling systems, and power supply options) would reduce accordingly. If fewer high-end GPUs are required, large-capacity information centers may downsize their investments in associated infrastructure, potentially affecting need for supporting innovations. This would put pressure on business that offer crucial components, most notably networking hardware, power systems, and cooling options.

Clear losers

Proprietary design suppliers

Why these innovations are positive: No clear argument.
Why these innovations are negative: The GenAI companies that have collected billions of dollars of financing for their exclusive models, such as OpenAI and Anthropic, stand to lose. Even if they establish and release more open models, this would still cut into the earnings flow as it stands today. Further, while some framed DeepSeek as a "side task of some quants" (quantitative experts), the release of DeepSeek's powerful V3 and then R1 designs proved far beyond that belief. The concern moving forward: What is the moat of exclusive model companies if cutting-edge models like DeepSeek's are getting released free of charge and become totally open and fine-tunable?
Our take: DeepSeek launched effective designs totally free (for regional deployment) or very cheap (their API is an order of magnitude more budget-friendly than equivalent designs). Companies like OpenAI, Anthropic, and Cohere will face increasingly strong competitors from players that launch complimentary and customizable advanced models, like Meta and DeepSeek.

Analyst takeaway and outlook

The emergence of DeepSeek R1 enhances a key pattern in the GenAI space: open-weight, cost-effective models are becoming viable rivals to proprietary alternatives. This shift challenges market assumptions and forces AI providers to reassess their worth propositions.

1. End users and GenAI application providers are the most significant winners.

Cheaper, premium models like R1 lower AI adoption expenses, benefiting both enterprises and forum.kepri.bawaslu.go.id customers. Startups such as Perplexity and Lovable, which develop applications on structure models, now have more options and can substantially lower API costs (e.g., R1's API is over 90% more affordable than OpenAI's o1 design).

2. Most professionals agree the stock exchange overreacted, but the innovation is genuine.

While significant AI stocks dropped dramatically after R1's release (e.g., NVIDIA and Microsoft down 18% and 7.5%, respectively), numerous analysts view this as an overreaction. However, DeepSeek R1 does mark an authentic advancement in expense performance and openness, setting a precedent for future competition.

3. The dish for constructing top-tier AI models is open, speeding up competition.

DeepSeek R1 has proven that releasing open weights and a detailed method is assisting success and accommodates a growing open-source neighborhood. The AI landscape is continuing to shift from a couple of dominant exclusive gamers to a more competitive market where brand-new entrants can construct on existing breakthroughs.

4. Proprietary AI service providers face increasing pressure.

Companies like OpenAI, Anthropic, and Cohere should now separate beyond raw design efficiency. What remains their competitive moat? Some may shift towards enterprise-specific options, while others might explore hybrid business designs.

5. AI infrastructure suppliers deal with combined prospects.

Cloud computing service providers like AWS and Microsoft Azure still gain from design training but face pressure as reasoning relocate to edge devices. Meanwhile, AI chipmakers like NVIDIA might see weaker demand for high-end GPUs if more models are trained with less resources.

6. The GenAI market remains on a strong growth path.

Despite disruptions, AI spending is anticipated to broaden. According to IoT Analytics' Generative AI Market Report 2025-2030, global costs on foundation models and platforms is forecasted to grow at a CAGR of 52% through 2030, driven by business adoption and continuous efficiency gains.

Final Thought:

DeepSeek R1 is not just a technical milestone-it signals a shift in the AI market's economics. The dish for constructing strong AI designs is now more commonly available, ensuring greater competition and faster innovation. While exclusive models should adjust, AI application companies and end-users stand to benefit a lot of.

Disclosure

Companies discussed in this article-along with their products-are used as examples to display market developments. No company paid or received favoritism in this article, and it is at the discretion of the expert to select which examples are used. IoT Analytics makes efforts to differ the companies and items mentioned to help shine attention to the many IoT and associated innovation market players.

It is worth keeping in mind that IoT Analytics might have business relationships with some business discussed in its posts, as some companies accredit IoT Analytics market research. However, for confidentiality, IoT Analytics can not disclose individual relationships. Please contact compliance@iot-analytics.com for any concerns or issues on this front.