Tenstorrent Raises $1.5B at $6.2B Valuation, Doubling Worth Since Late 2024
Chip startup backed by Trace Capital and Innovation X plans to scale inference hardware for open-weight models, rejecting acquisition offers from Intel and Qualcomm.
Last verified:
Tenstorrent’s $6.2B Valuation Marks Steep Climb From 2024
Tenstorrent, an inference-chip startup led by microprocessor engineer Jim Keller, secured $1.5 billion in Series C funding at a $6.2 billion post-money valuation, according to TechCrunch AI. The round, led by Trace Capital (a new venture vehicle backed by Quiet Capital co-founder Ben Mahdavi) and Innovation X, more than doubled the company’s valuation from the $2.6 billion it achieved in late 2024—a 138% increase in roughly 22 months.
Co-investors in the round included Cadence Ventures, Eclipse, Fidelity, SanDisk, and T. Rowe Price, reflecting growing institutional appetite for semiconductor alternatives to Nvidia’s dominance in AI accelerators.
Why Independence Over Acquisition
The funding announcement arrives roughly five months after Tenstorrent reportedly fielded acquisition approaches from both Intel and Qualcomm at valuations as high as $5 billion. Keller’s decision to pursue venture funding rather than accept either bid signals confidence in the company’s standalone trajectory.
According to TechCrunch AI, Keller attributed the shift toward remaining independent to a technical milestone: his Galaxy servers had demonstrated the ability to efficiently run DeepSeek models. This proof-of-concept appears to have vindicated his thesis that open-weights AI inference—not proprietary models—represents the market opportunity justifying a higher valuation than the acquisition offers implied.
Scaling Chips for Open-Weights and Video
The capital infusion is earmarked for scaling production of inference silicon tailored specifically to serving open-weight AI models and video-generation workloads. This positioning places Tenstorrent within a broader cohort of startups attempting to erode Nvidia’s grip on AI chip supply.
Etched, a rival inference-chip startup, was fielding investment proposals at valuations exceeding $40 billion as of mid-October 2026, per TechCrunch’s earlier reporting that same month—nearly seven times Tenstorrent’s current valuation and underscoring the divergent capital appetites across the AI chip stack.
Why This Matters
The $1.5 billion raise validates a strategic bet on open-weights model inference as a defensible market segment distinct from training acceleration. For teams evaluating alternative silicon to Nvidia for serving open-source models like DeepSeek and Meta’s Llama family in production, Tenstorrent’s Galaxy hardware moves from speculative to operationally proven. The company’s rejection of acquisition at lower valuations suggests founders believe the inference-serving market will command premium economics as model weights become commoditized and inference volume grows. Whether this thesis holds depends on Galaxy servers achieving cost-per-inference parity or advantage over entrenched competitors—a claim that will face independent scrutiny as deployment data accumulates.
Frequently Asked Questions
Why did Tenstorrent reject acquisition offers to raise venture capital instead?
CEO Jim Keller stated the company chose independence after proving its Galaxy servers could run DeepSeek models efficiently, signaling stronger market validation than the $5 billion acquisition proposals from Intel and Qualcomm.
What will Tenstorrent use the $1.5 billion for?
The capital will fund scaling production of inference chips optimized for open-weight AI model serving and video generation workloads.
How does Tenstorrent's valuation compare to other AI chip startups?
At $6.2 billion, Tenstorrent trails Etched AI, which was fielding investment offers at $40 billion or higher valuations as of mid-October 2026.