Who is Tuhin Srivastava?
Tuhin Srivastava is a technology executive, likely leading an AI infrastructure or inference company. The material tracks his views on the inference market, customer retention, and compute scarcity.
Track record
- May 2026 - Srivastava said none of his top 30 customers have ever churned, with annual net dollar retention around 400%, per his own statement.
- May 2026 - A host stated that Srivastava’s company is expected to exceed $1 billion in revenue that year.
- May 2026 - Srivastava described inference as the last market, even if AGI is achieved, per his own remark.
- May 2026 - Srivastava noted that no one runs vanilla open-source weights, implying custom optimization is standard, per his own statement.
- May 2026 - Srivastava said by inference count, 99% of usage comes from custom or fine-tuned models, per his own estimate.
- May 2026 - Srivastava argued there will never be enough compute to extract the desired value from language models in the next 5–10 years, per his own forecast.
On the record
What named speakers have said about Tuhin Srivastava.
Baseten reports 400% annual net dollar retention, with zero churn among its top 30 customers.
“None of our top 30 customers have ever turned. You know, we're talking like 400% annual NDR.”Tuhin Srivastava · 1 May 2026
Baseten projects more than $1 billion in revenue for 2026, according to the host.
“I think I can say you're expecting to do more than a billion dollars in revenue this year.”
Tuhin Srivastava frames inference as the permanent, irreducible market in AI, even beyond AGI.
“It is the last market, right? Like even if there's AGI, all that's left is inference.”Tuhin Srivastava · 1 May 2026
Tuhin Srivastava on the universality of model customization among Baseten customers.
“No one is just running the vanilla open source weights.”Tuhin Srivastava · 1 May 2026
Srivastava argues that inference is the enduring core of the AI market because even a post-AGI world still requires compute to run models and extract value from them.
“It is the last market, right? Like even if there's AGI, all that's left is inference.”Tuhin Srivastava · 1 May 2026
By inference count, roughly 99% of AI inference is driven by application-layer companies rather than direct enterprise adoption.
“I think if you look by inference count, it'd be 99% the former.”Tuhin Srivastava · 1 May 2026
Srivastava contends that available compute will be fundamentally insufficient to capture the full value of LLMs over the next 5 to 10 years, implying a persistent and worsening supply constraint.
“There's no world in which there's enough compute to you know, get the amount of the amount of value that we want to get out of L M's in the next 5 to 10 years.”Tuhin Srivastava · 1 May 2026
Srivastava argues no competitor can realistically challenge Nvidia in the next few years due to entrenched supply chain and ecosystem advantages, pushing back on narratives of imminent GPU market disruption.
“I'm not saying it won't happen like the short-term like in the next couple years how anyone can be able to compete with.”Tuhin Srivastava · 1 May 2026