Jensen Huang said in February 2026 that NVIDIA's Grace Blackwell with NVLink is "the king of inference today," delivering "an order-of-magnitude lower cost per token," without stating what that cost is compared with.
Resolved true The record shows it happened or was so.
Factual: Statements about the present or the past, checked against the record: did they hold up?
Said for an organisation they lead or work for; it counts on both records.
What would count as true or false
This claim is about what Huang said. True: a record (transcript, video, NVIDIA press release or earnings remarks) from around February 2026 shows Huang saying that Grace Blackwell with NVLink is 'the king of inference today' with 'an order-of-magnitude lower cost per token', and that the statement gave no explicit point of comparison. Mostly true: the wording is substantially the same with minor differences, or the date is slightly off. Mixed: he made the statement but did state a point of comparison (for example versus Hopper) in the same passage. Mostly false: he made a materially different statement, for example without the order-of-magnitude cost claim. False: no such statement by Huang can be found and it was attributed to someone else, or the record shows he said something different.
Set before the research began.
Evidence (2)
-
E1 NVIDIA's financial results show record quarterly revenue of ... TweakTown · 25 February 2026
“Grace Blackwell with NVLink is the king of inference today, delivering an order-of-magnitude lower cost per token, and Vera Rubin will extend that leadership even further.”
-
E2 NVIDIA Announces Financial Results for Fourth Quarter and Fiscal 2026 NVIDIA · 25 February 2026 Own record
“Revealed that NVIDIA Blackwell Ultra delivers up to 50x better performance and 35x lower cost for agentic AI compared with the NVIDIA Hopper platform, according to new SemiAnalysis InferenceX benchmark results.”
Reasoning
E2, an independent report dated 2026-02-25 on NVIDIA's fourth-quarter results, gives the exact wording: "Grace Blackwell with NVLink is the king of inference today, delivering an order-of-magnitude lower cost per token." The sentence states no point of comparison for the cost claim. The surrounding text of the claim attributes the sentence to Huang, and E3 is NVIDIA's results release from the same day [E3]. The comparison with Hopper quoted from E3 is about Blackwell Ultra and sits in a separate bullet, not in this statement [E3]. One limitation is that the quoted text in E2 does not itself name Huang; the attribution comes from the claim's surrounding text.
Cite this claim
Jensen Huang, “Jensen Huang said in February 2026 that NVIDIA's Grace Blackwell with NVLink is "the king of inference today," delivering "an order-of-magnitude lower cost per token," without stating what that cost is compared with.” (said 25 February 2026). Verdict: resolved true. Reliability API, claim c_a1dfbc38c42b, method 0.1.0+03d23c0e, dataset release v1.1.2. https://epistacklabs.com/claims/c_a1dfbc38c42b
This claim is in dataset release v1.1.2, with the same verdict.
Reply to this claim
More factual claims by Jensen Huang
-
Mostly true
“we invented a technology called a GPU, and basically a GPU does parallel processing tasks”
-
Resolved false
“Today, NVIDIA’s token cost is the lowest in the world, not by 10%, by X factors, orders of magnitude, all because we did extreme co-design”
-
Mostly true
“It’s the first one to be PCI Express Gen 6.”
-
Mixed
“We have demonstrated that NVFP4 can achieve inference without loss of precision, while delivering significant performance and energy efficiency improvements, and it is equally suitable for training.”
-
Mixed
“Over the past two years, the computational demands for inference have increased approximately 10,000 times, while usage has increased approximately 100 times.”
Verdicts are made automatically; how verdicts are made.