43:I[185,["/_next/static/chunks/0sxns9k8ywops.js","/_next/static/chunks/1z5eijs5qdv02.js","/_next/static/chunks/2eylv_sshoiot.js","/_next/static/chunks/3owl5o1y-37u7.js","/_next/static/chunks/0agcep-t686m_.js","/_next/static/chunks/09e73exje0qtu.js","/_next/static/chunks/1nsk8jrr9fqe6.js"],"SendEmailButton"] 45:I[63625,["/_next/static/chunks/0sxns9k8ywops.js","/_next/static/chunks/1z5eijs5qdv02.js","/_next/static/chunks/2eylv_sshoiot.js","/_next/static/chunks/3owl5o1y-37u7.js","/_next/static/chunks/0agcep-t686m_.js","/_next/static/chunks/09e73exje0qtu.js","/_next/static/chunks/1nsk8jrr9fqe6.js"],"AddToLibraryButton"] 46:I[29721,["/_next/static/chunks/0sxns9k8ywops.js","/_next/static/chunks/1z5eijs5qdv02.js","/_next/static/chunks/2eylv_sshoiot.js","/_next/static/chunks/3owl5o1y-37u7.js","/_next/static/chunks/0agcep-t686m_.js","/_next/static/chunks/09e73exje0qtu.js","/_next/static/chunks/1nsk8jrr9fqe6.js"],"ShareButton"] 47:I[58016,["/_next/static/chunks/0sxns9k8ywops.js","/_next/static/chunks/1z5eijs5qdv02.js","/_next/static/chunks/2eylv_sshoiot.js","/_next/static/chunks/3owl5o1y-37u7.js","/_next/static/chunks/0agcep-t686m_.js","/_next/static/chunks/09e73exje0qtu.js","/_next/static/chunks/1nsk8jrr9fqe6.js"],"ProcessEpisodeButton"] 48:I[45269,["/_next/static/chunks/0sxns9k8ywops.js","/_next/static/chunks/1z5eijs5qdv02.js","/_next/static/chunks/2eylv_sshoiot.js","/_next/static/chunks/3owl5o1y-37u7.js","/_next/static/chunks/0agcep-t686m_.js","/_next/static/chunks/09e73exje0qtu.js","/_next/static/chunks/1nsk8jrr9fqe6.js"],"SignupCTA"] 3b:["$","time",null,{"dateTime":"2026-06-30T13:18:00.000Z","children":"June 30, 2026"}] 3c:["$","span",null,{"aria-hidden":true,"children":"·"}] 3d:["$","span",null,{"children":"00:07:44"}] 44:T439,

Angelos Perivolaropoulos, a research engineer at ElevenLabs, on turning GPU scarcity into an inference-engineering problem: how to serve far more users on the same hardware, from batching to frontier architecture changes. Recorded at RAAIS 2026.

00:00 Introduction: ElevenLabs and the GPU squeeze

00:38 The question: how to scale when you can't add capacity

01:11 About Angelos: Scribe, speech-to-text and text-to-speech

01:56 GPU scarcity meets exponential demand

02:44 What a token actually costs: compute vs memory bandwidth

03:38 Prefill, decode and the KV cache

05:53 Batching and continuous batching (1 → 15 users/GPU)

08:37 FP8 quantization and quantize-aware training (→ 20)

11:29 Speculative decoding and multi-token prediction (→ 28)

15:13 Compressing the KV cache: TurboQuant and distillation (→ 70)

17:27 Frontier architectures: MLA, linear attention, state-space (→ 140)

20:39 Trade-offs: nothing is free

22:03 Q&A: papers vs production, token subsidies, TTS evals