Tenstorrent Raises $1.5 Billion at a $6.2 Billion Valuation After Proving Galaxy Servers Can Run DeepSeek Models
The company has chosen to move forward independently and will use the funds to expand chips for serving open-weight AI models (models whose weights are publicly available for download and further use), as well as video generation and free credits.
In brief: Tenstorrent raised $1.5 billion at a $6.2 billion valuation after successfully running DeepSeek on Galaxy servers. The company will use the funds to expand chips for open-weight AI and video generation while remaining independent and challenging Nvidia’s position.
If I were to interpret this, what makes the news interesting is that Tenstorrent chose not to sell the company, but instead decided to continue independently. This reflects the company’s belief that there is still room for new players in the AI chip market, even though Nvidia is the market leader. Tenstorrent remains the smaller player that must continue proving its performance and cost advantages.
What This News Reveals About Tenstorrent
The increase from a $2.6 billion valuation in late 2024 to $6.2 billion reflects that investors still believe in alternatives to Nvidia, particularly in the market for chips supporting open-weight AI models.
Tenstorrent has also chosen to remain independent and is preparing to expand chips for AI model serving and video generation. This gives the company a clearer role among AI hardware startups.
When AI Companies Need to Reduce Their Dependence on Nvidia
Serving open-weight AI models requires chips that are well suited to the workload. Nvidia remains dominant in the market, prompting many AI hardware startups to try to claim a share of it.
Tenstorrent has chosen to focus on chips for serving open-weight models and video workloads, while continuing to raise funds and operate as an independent company. This approach reflects a growing market search for alternatives to Nvidia.
Where Tenstorrent Fits in the AI Chip Market
Tenstorrent is a 10-year-old chip startup focused on inference (the processing required for an AI model to generate answers or outputs) and positioned as an Nvidia challenger. Its key strength is developing chips specifically designed to support open-weight AI models.
The fact that Galaxy servers can run DeepSeek models has led the company to remain independent and raise additional funding instead of selling to a major company such as Intel or Qualcomm. Maintaining its independence gives Tenstorrent room to control its chip strategy and continue expanding the market on its own terms.
If the AI chip market were a racecourse, Nvidia would be the established leader with a strong ecosystem, while Tenstorrent would be the smaller player trying to prove that it has its own distinct path, particularly in inference and open-weight AI.
From a $2.6 Billion Valuation to $6.2 Billion
Tenstorrent’s valuation has risen from $2.6 billion in late 2024 to $6.2 billion in the latest round, alongside $1.5 billion in funding. Key factors include Galaxy servers’ ability to run DeepSeek models and the company’s decision to continue independently.
| Factor | Late 2024 | Latest funding round |
|---|---|---|
| Company valuation | $2.6 billion | $6.2 billion |
| Funding raised | Not specified in the source | $1.5 billion |
| Time elapsed | Reference point in late 2024 | 5 months after reports of acquisition interest |
| Relevant factors | Valuation before the latest round | Galaxy servers can run DeepSeek, and funding was raised to expand chips for open-weight AI |
This round therefore reflects confidence in the plan to expand chips for open-weight AI and video generation, rather than growth based solely on valuation figures.
What Will the New Funding Be Used For?
The new funding will help Tenstorrent expand chips for serving open-weight AI models, in particular. This type of workload suits teams that want to serve models on their own servers.
The company has already demonstrated that Galaxy servers can run DeepSeek models. Another focus is developing chips to support video generation, a workload that requires processing systems designed specifically for serving models. The funding will therefore be used to increase the company’s capabilities and push its products further into the market.
How Does Tenstorrent Compare with Other Players?
Tenstorrent focuses on chips for inference and serving open-weight AI models, while Nvidia is regarded as the market leader that AI hardware companies are trying to challenge. Etched is an AI chip manufacturer reportedly receiving investment offers valued at $40 billion or more.
| Factor | Tenstorrent | Nvidia | Etched |
|---|---|---|---|
| Market role | Chips for inference and open-weight AI models | Dominant AI chip company | AI chip manufacturer |
| Competition | Trying to claim market share from Nvidia | Primary target for new AI chip companies | Among the companies challenging Nvidia |
| Valuation mentioned | $6.2 billion | No information in the source | $40 billion or more |
| Technical details | The source only states that Galaxy servers can run DeepSeek models | Insufficient details for comparison | Insufficient details for comparison |
Strengths and Limitations Based on the Available Information
Pros
- +Raised $1.5 billion at a $6.2 billion valuation
- +Galaxy servers can run DeepSeek models
- +Plans to expand chips for open-weight AI and video generation
Cons
- −No detailed chip specifications are available
- −No performance, pricing, or competitor benchmark data is available
- −Value for real-world use cannot yet be assessed
Costs That Cannot Yet Be Evaluated
The news does not disclose chip prices, Galaxy server costs, system usage fees, or the expenses involved in real-world deployment. As a result, the total cost cannot yet be calculated from the available information.
Pros
- +The new funding will be used to expand chips for open-weight AI
- +The scope of the information disclosed in the news is clear
Cons
- −No chip prices or server costs
- −No system usage fees or real-world installation and deployment costs
What to Watch After This Funding Round
The key question is how far Tenstorrent can build on Galaxy servers’ ability to run DeepSeek models and scale that capability into widespread real-world use. It will also be important to see whether the company discloses specifications, performance, pricing, and total costs compared with Nvidia.
If I were to give a recommendation, I would view Tenstorrent as a challenger showing positive signs, but it is still too early to conclude that it can replace Nvidia until more verifiable performance and cost data becomes available.