The American AI market was recently rattled by the emergence of a Chinese competitor that’s cost-efficient and matches the performance of OpenAI’s o1 model on several math and reasoning metrics.
The new AI model, created by Hangzhou-based startup DeepSeek, has supposedly surpassed Meta as the leading purveyor of open-source AI tools. The company’s R1 model is alleged to cost just $6 million to train– a fraction of what it costs companies like NVIDIA and Microsoft to train their models– and its most powerful versions cost approximately 95 percent less than OpenAI and its competitors.
While many are unsure about DeepSeek’s claims regarding how much the company has spent and how many advanced chips it deployed to create its model, few dispute the AI model’s game-changing capabilities.
“DeepSeeks’ ability to produce results comparable to Western AI giants using non-premium chips has drawn enormous international interest– with interest possibly further increased by recent news of Chinese apps such as the TikTok ban and REDnote migration,” said Ted Miracco, CEO of Approov. “Its affordability and adaptability are clear competitive advantages, while today, OpenAI maintains leadership in innovation and global influence. This cost advantage opens the door to unmetered and pervasive access to AI, which is to be both exciting and highly disruptive.”
What makes DeepSeek’s AI model so intriguing?
DeepSeek’s new AI model’s rapid progress and minimal investment sent shockwaves through the industry, causing IT stocks to tumble and AI strategies to be rethought.
According to Phillip Walker, Customer Advocate CEO of Network Solutions Provider USA, DeepSeek’s model was accelerated in development by learning from past AI pitfalls and challenges that other companies have endured.
“What I think is absolutely compelling is how fast they caught the frontier,” Walker said. “They’ve taken all the learning expertise or pitfalls and challenges that every other AI machine has had and accelerated through it. So, they haven’t had to have a major investment.”
From a business standpoint, organizations may be taking a look at how they’re currently developing AI and whether they need to reorient their approaches. According to The Information, a tech news site, Meta has set up four “war rooms” to analyze DeepSeek’s models, seeking to find out how the Chinese tech startup trained a model so cheaply and to use the insights to improve their own open source Llama models.
DeepSeek has published the information on their AI model and one can test their models and APIs to see what they’ve accomplished.
“They’re not hiding the infrastructure framework of their models. Is there an opportunity to look at what they did and use it to accelerate your space? Absolutely, yes,” said Walker. “The fact that they’ve used that large model to make their small models more efficient– a lot of companies can now adapt that practice and use that to accelerate their growth very fast.”
What does this mean for the Channel?
Currently, this new development does not mean a whole lot for the channel.
As it stands right now, the channel is behind the curve on AI developments and has so far not had the opportunity to catch up. There’s still a gap from a skills standpoint of moving from a digital transformation company to a digital AI company. While using AI does accelerate that process, having the skills to develop and lead channel organizations is not there yet.
“We have not identified the gaps that are necessary to be filled,” said Walker. “In the past, your IT solution provider’s value has been filling the gaps, allowing for customer success and deploying technology when businesses do not have that. The business impact, the implications, the security– those things are not there yet from a channel standpoint, but we will get there.”
Concerns surround DeepSeek’s model
The origins of DeepSeek’s AI model have naturally sparked debates over national security. U.S. officials have raised concerns over the use of this technology and its access to U.S. users.
Cloud and network security company, Wiz, saw its research team uncover an exposed DeepSeek database leaking sensitive information, including chat history. A publicly accessible database belonging to DeepSeek allowed full control over database operations, exposing over a million lines of log streams and highly sensitive information, such as chat history, secret keys, and backend details.
“This database contained a significant volume of chat history, backend data, and sensitive information, including log streams, API Secrets, and operational details,” Wiz’s research stated. “More critically, the exposure allowed for full database control and potential privilege escalation within the DeepSeek environment, without any authentication or defense mechanism to the outside world.”
House Speaker Mike Johnson, R-La., claimed that DeepSeek is “a serious threat” that should be dealt with in an appropriate manner.
Additionally, Sen. Mark Warner, D-Va., defended the existing export controls that prevent advanced U.S. chip technology from being acquired by China, while suggesting that more regulation might be required.
“While I think there’s more to learn about DeepSeek’s development activities, what’s in the public record reveals that the PRC (People’s Republic of China) continues to prioritize advancement in AI and that export control alone will not stymie their efforts,” said Warner.
Further, OpenAI has since uncovered evidence that its proprietary models were used by DeepSeek to train their AI model, potentially violating OpenAI’s terms of service. DeepSeek used a technique known as “distillation,” which is where developers use outputs from larger AI models to train smaller ones. This is common practice in AI development, but OpenAI claims DeepSeek took the practice too far in developing their rival model.
“The issue is when you take it out of the platform and are doing it to create your own model for your own purposes,” an OpenAI source told the Financial Times.
Director of Information Security and Engagement at the National Cybersecurity Alliance (NCA) Cliff Steinhauer offered that the path forward for AI requires balancing innovation with robust data protection and security measures.
“The technology advancements demonstrated by DeepSeek raise important considerations about data governance and privacy frameworks across different regulatory environments,” Steinhauer said. “Chinese AI companies operate under distinct requirements that give their government broad access to user data and intellectual property. This creates unique challenges when considering the use of these AI systems by international users, particularly for processing sensitive or proprietary information.”
While the emergence of this new player in the world of AI impacted the stock prices of companies like NVIDIA significantly, chipmakers will still have time to adjust to the potentially new landscape of AI. Countries like Russia and Israel could be poised to make a significant impact in the AI market as well, along with tech giants like Apple– a company that has kept its AI plans close to the vest.
“It’s shown the path for the channel that there is a level playing field to have your own AI applications and large language models,” Walker said. “If you’re in the channel and you’re not doing large language models, you’re not touching machine learning or data sets. This is the open gate to do that because the major enablers of AI are going to be data. The second piece is going to be: what does the solution provider look like and how is that run? Because if you’re using the capabilities, it’s going to make a lot of your processes faster, and you’re going to be able to keep up with businesses, but you need to get ahead and that’s where we’re close to the front, but just not there yet.”
While the emergence of DeepSeek has big implications across the industry, other major players continue to make AI-related news. Read more about ServiceNow’s AI partnerships with several tech giants.