Tenstorrent Raises $1.5 Billion at a $6.2 Billion Valuation: Can It Challenge NVIDIA?
Tenstorrent raised $1.5 billion at a $6.2 billion valuation. Here is what it sells, who invested, how it compares with NVIDIA, and what to watch next.
The short answer
Tenstorrent closed a $1.5 billion Series E at a $6.2 billion post-money valuation on October 19, 2026. The money confirms investor confidence and pays for scaling Galaxy servers; it does not prove the company can match NVIDIA on performance, software or adoption.
Tenstorrent raised $1.5 billion at a $6.2 billion valuation. Here is what it sells, who invested, how it compares with NVIDIA, and what to watch next.
01 / The story
What happened: Tenstorrent's $1.5 billion Series E
Tenstorrent, the AI computing company led by chip architect Jim Keller, said on October 19, 2026 that it closed $1.5 billion in Series E funding at a $6.2 billion post-money valuation [1]. Reuters reported that this more than doubles the valuation from the company's December 2024 round [2].
The news matters because most AI computing today runs on NVIDIA hardware. A large new round gives one of the better-known challengers more money to build and ship systems. It is a signal of investor belief, not proof that Tenstorrent will take market share.
02 / The story
What is Tenstorrent?
Tenstorrent builds AI processors and complete systems based on RISC-V, an open instruction set that companies can use to design processors without paying architecture license fees [1] [2]. If RISC-V is new to you, our explainer on [RISC-V](/articles/risc-v-explained) covers the basics.
The company says it sells AI compute "at every level of the stack": a single card or workstation, racks, large clusters, and custom intellectual property (IP) [1]. That last part is important. Besides selling hardware, Tenstorrent licenses its TT-Ascalon RISC-V CPU design and Tensix AI cores to chip designers, and names Samsung and LG as licensees [1].
Tenstorrent describes its software stack as open source [1]. That does not mean every Tenstorrent chip design, license or product is open. The licensed CPU and AI core designs are commercial IP sold to customers.
03 / The story
Who invested, and how will the money be used?
The round was led by Trace Capital and Innovation X. Named participants include ARK Invest, Fidelity Management & Research Company, Morgan Stanley, Sandisk, T. Rowe Price, Tether Investments, Eclipse, Maverick Silicon, Protagonist, Virtu Financial and XTX Markets, among others [1].
Tenstorrent says it will use the capital to deploy its Galaxy servers at scale, keep serving open-weight models for language and video generation, and advance the AI and CPU IP it licenses [1]. Reuters describes Galaxy servers as systems combining dozens of the company's AI chips [2]. These are stated plans, not completed results.
A quick money glossary helps here. The funding amount ($1.5 billion) is cash investors put in. The post-money valuation ($6.2 billion) is the price investors agreed the whole company is worth after that cash arrives. Neither number is revenue, profit, or a public stock market value.
04 / The story
Tenstorrent vs NVIDIA: a balanced comparison
The table compares business focus and stated offerings, not measured speed. Performance depends on the model, memory needs, software, system setup, energy use and total cost of ownership.
| Dimension | Tenstorrent | NVIDIA |
|---|---|---|
| Main role | AI compute company selling systems and licensing CPU and AI core IP [1] | Dominant supplier of AI GPUs, systems and software for data centers and PCs [5] |
| Hardware | RISC-V-based AI processors, cards, workstations and Galaxy servers [1] | GPU-based systems, including RTX Spark PCs and DGX workstations [5] |
| Software | Open-source software stack, per the company [1] | CUDA platform and libraries such as CUDA-X [4] [5] |
| Deployment | Local workstations, on-premises racks and clusters [1] | PCs, workstations, data centers and many cloud partners [4] |
| Openness | Open software stack; licensed IP is commercial [1] | Proprietary hardware and CUDA; some models and tools released openly [4] |
| Watch for | Shipping volume, framework support, independent benchmarks | Supply, pricing and how well its open tools fit your stack |
05 / The story
What open-weight AI means for hardware demand
Model weights are the learned numbers that make a model work. When a company releases them, anyone allowed by the license can run the model on their own equipment. That creates demand for hardware outside the big model providers' clouds, which is the market Tenstorrent says it serves [1].
Open weights do not make the training data, training process, hardware or every software tool open. For hardware buyers, three questions matter most: how efficiently the chip runs the model, how flexibly it can be deployed, and whether the software runs your model without rewrites. Our guide to [local open-source LLMs](/articles/local-open-source-llms) explains the model side.
06 / The story
Why enterprises may care
These examples are illustrative, not reported Tenstorrent deployments.
- AI infrastructure teams: a second hardware option can help with supply and price negotiations, if it runs their models well.
- Companies building internal assistants: on-premises inference can keep data on owned systems, though it still needs security work.
- Cloud and inference providers: another supplier can diversify hardware, at the cost of supporting another software stack.
- Chip designers: licensing CPU and AI cores can shorten the path to custom silicon, as Tenstorrent says Samsung and LG do [1].
07 / The story
What could limit Tenstorrent's growth
These are factors to evaluate, not confirmed problems. Many developers already know NVIDIA's tools, so software maturity and framework compatibility decide how easy a switch is. Manufacturing at scale, delivery times, customer support and system integration also matter. Performance can vary widely between workloads, and moving an existing stack carries its own cost. NVIDIA, other chip startups and cloud providers' in-house chips all compete for the same buyers. Our look at [who pays for AI data center investment](/articles/ai-data-center-investment-who-pays) covers the wider spending picture.
08 / The story
What happens next: milestones to watch
The funding becomes meaningful only if it turns into measurable results. Look for:
- Galaxy servers shipping in volume and becoming easy to buy.
- Named commercial deployments confirmed by customers.
- Support for popular AI frameworks and models.
- Independent benchmarks on the same workloads as rival systems.
- New IP licensing deals and partnerships.
09 / The story
Conclusion
The round confirms that large investors will pay a much higher price for Tenstorrent than in 2024, and it funds a clear plan: more Galaxy servers and more licensed IP [1] [2]. It does not show that Tenstorrent beats NVIDIA. If you are evaluating alternative AI hardware, test your own models on it and compare cost per finished task before believing anyone's headline.
A closer look
Evidence and limits
Funding size, valuation, investors and stated uses come from Tenstorrent's own announcement, checked on October 19, 2026, and match Reuters' report the same day. Investor quotes are attributed to the people who made them.
We found no independent, like-for-like benchmark comparing Tenstorrent and NVIDIA systems on the same workloads, so this article does not rank them. Revenue, customer counts and production volumes were not disclosed and are not estimated here.
Put it to work
Before testing alternative AI hardware
Run these checks on your own workloads, not on vendor slides.
0 of 6 considered
These questions are editorial guidance, not a product score or verified assessment.
Keep learning
Further reading
- What Is RISC-V? The Open Processor Architecture Explained
Qlauson · Emerging Technology
- The AI Data Center Boom: Who Pays If the Investment Doesn't Pay Off?
Qlauson · Cloud and Infrastructure
- How to Run Open Source LLMs Locally (2026)
Qlauson · Software
- AI Model Costs: When Does Self-Hosting Break Even for Business?
Qlauson · Cloud and Infrastructure
- Tenstorrent: Tenstorrent Announces a $1.5B Series E at a $6.2B valuation
External resource · Official announcement: amount, valuation, investors, plans and licensees, Oct 19, 2026. Opens in a new tab.
- Reuters: AI chip startup Tenstorrent more than doubles its valuation
Primary documentation · Independent report on the round and the December 2024 valuation. Checked Oct 19, 2026. Opens in a new tab.
- Tenstorrent Trademark Guidelines
Primary documentation · Logo use requires review; our cover uses plain name text instead. Checked Oct 19, 2026. Opens in a new tab.
- NVIDIA: Enterprise Software Leaders Build AI Agents With NVIDIA
Primary documentation · NVIDIA open models, CUDA-X and partner ecosystem, June 1, 2026. Checked Oct 19, 2026. Opens in a new tab.
- NVIDIA Blog: Local AI, RTX Spark and the Microsoft Windows event
Primary documentation · NVIDIA PC and workstation hardware and the CUDA platform. Checked Oct 19, 2026. Opens in a new tab.
Read next
DeepSeek’s New AI Model: What the Latest Release Means for Developers
DeepSeek’s October 17 release brings MIT-licensed weights and web APIs. We examine its benchmark limits, API prices and practical developer trade-offs.
Continue readingNewsletter
Get new Qlauson stories by email.
An occasional email when we publish a new guide or update an important one. No unrelated promotions, and you can unsubscribe from any email.
