Ashish Vaswani, the Essential AI co-founder and one of the authors of the Transformer paper, is reportedly moving to NVIDIA (@nvidia) as part of an acquihire of the Essential AI team into NVIDIA's Nemotron group.
Sharon Goldman on X The post says NVIDIA acquihired Essential AI's team, including Vaswani, and added the group to Nemotron. It does not state when the hires closed, how many Essential AI employees joined NVIDIA, whether NVIDIA bought Essential AI's assets or IP, or what consideration was paid.
That distinction matters. This is not yet a confirmed acquisition of Essential AI as a company. The cleanest read is narrower: NVIDIA has reportedly hired a team built around one of the researchers whose 2017 work made the modern large language model era possible.
Vaswani is the reason the move matters. He was listed as the first author of the 2017 paper "Attention Is All You Need", alongside Noam Shazeer, Niki Parmar, Jakob Uszkoreit, Llion Jones, Aidan N. Gomez, Lukasz Kaiser, and Illia Polosukhin. The paper introduced the Transformer architecture, replacing recurrent and convolutional sequence models with an attention-based design that could be parallelized more efficiently. That technical shift became the substrate for the current generation of language models.
After Google Brain, Vaswani co-founded Adept AI and later [Essential AI](https://www.essential.ai/?ref=runtimewire).
If accurate, placing Essential AI talent inside Nemotron would strengthen NVIDIA's in-house model research alongside its dominance in compute. It would also give NVIDIA people who have lived through the shift from academic architecture work to venture-backed model company building.
NVIDIA has also announced the Nemotron Coalition to advance open frontier models, signaling a push up the stack into models, datasets, and training recipes that shape how developers use its hardware.
For founders, the Essential AI report is a reminder of how the AI infrastructure market is consolidating around people as much as products. A model startup's IP matters. Its benchmarks matter. Its customers matter. But in frontier AI, the highest-value asset is often the small group of researchers and engineers who know how to make the next training run work. The open questions are still material. The report does not disclose how many Essential AI employees are joining NVIDIA, whether other Essential AI leaders are part of the move, whether Essential AI continues independently, or whether investors received any return beyond employment packages for the team. It also does not establish when the transaction occurred. Until those details are confirmed, this should be read as a reported talent move, not a full corporate acquisition.
Even with those caveats, the direction is clear. NVIDIA is not waiting for model companies to define the workload that drives demand for NVIDIA systems. NVIDIA is building, backing, partnering with and, when useful, hiring the people who define that workload itself.