While Cerebras builds custom AI chips, French startup Kog squeezes performance out of the GPUs enterprises already own. By rewriting low-level software for existing datacenter hardware, Kog eliminates the latency bottlenecks that slow AI workflows.

Challenging the Necessity of Custom AI Silicon

The AI industry has pivoted to purpose-built chips for inference. Kog's CEO, Gaël Delalleau, says the belief that standard GPUs can’t handle decoding is wrong. Modern GPUs—Nvidia H200, AMD MI300X—offer memory bandwidth that current software leaves idle.

Kog’s Kog Inference Engine (KIE) targets "extremely fast single-request decoding," a must-have for real-time apps. In a recent demo the company hit 3,000 tokens per second (TPS) with Laneformer 2B, an open-source 2-billion-parameter model tuned for their stack. The demo uses a small model, but Kog plans to extend those gains to much larger LLMs.

A "Hacker" Approach to GPU Engineering

Unlike hardware-agnostic providers such as ZML, Kog digs deep. The team reverse-engineers GPUs at the assembly and binary levels, treating the silicon as a set of physical laws to master.

That low-level focus extracts every ounce of efficiency, but it costs time. With a lean crew of 11 engineers, Kog spends weeks or months dissecting each new GPU architecture before adding support. The method delivers top-tier performance but limits how quickly Kog can cover new hardware.

Targeting High-Value AI Use Cases

Kog focuses on sectors where latency equals lost revenue:

  • Software engineering: Tools like Claude Code stall for minutes. Kog aims to turn "waiting hours" into near-instant results.
  • Generative app/game design: Prompt-to-app pipelines need rapid iteration to keep users engaged and revenue flowing.

The company’s next milestone is a 10× speedup on its first major large-scale model. Backed by Scaleway, Bpifrance and the French Tech 2030 program, Kog positions itself as a key player in Europe’s AI sovereignty drive.

Key Takeaways

  • Optimization over hardware: Kog pushes the memory bandwidth of Nvidia H200 and AMD MI300X GPUs to the limit with extreme low-level software engineering.
  • Breaking the latency barrier: The startup targets a 30× boost in LLM inference speed for real-time professional workflows such as automated coding and generative design.
  • Deep-level engineering: By adopting a "hacker" mindset, Kog reverse-engineers GPU assembly and binary code to achieve performance that standard stacks cannot reach.

Kog, a French startup, says its Kog Inference Engine aims to run large-language-model (LLM) decoding up to 30 times faster on the same Nvidia H200 or AMD MI300X GPUs that data-center operators already own.

Why the push for speed matters now

LLM inference now throttles products like code-completion assistants, on-the-fly content generators, and interactive game-design tools. In those settings, the gap between a user’s prompt and the model’s reply directly impacts productivity and revenue. High-level APIs such as CUDA leave a large slice of GPU memory bandwidth idle, especially during token-by-token decoding, which dominates real-time use cases.

Kog’s low-level answer

Kog skips conventional software layers and talks straight to the silicon. Its engineers reverse-engineer the GPU at the assembly level, treating the hardware as a set of constraints to obey rather than a black box to abstract. The result is a custom execution path that keeps data flowing through the GPU’s memory pipes at near-full capacity.

In a recent demo the company ran Laneformer 2B—a 2-billion-parameter open-source model tuned for its stack—and reported a high token throughput. The model is modest compared with larger commercial systems, but the speed gain shows what a purpose-built software stack can extract from the same hardware.

The engineering trade-off

The upside comes with a steep cost curve. Kog’s 11-engineer team spends weeks or months dissecting each new GPU architecture before it can be supported. That depth yields high performance on the targeted cards, but it also means Kog cannot instantly add support for every new GPU that hits the market. Customers benefit only if their fleets include the Nvidia H200 or AMD MI300X GPUs Kog has already optimized.

Who stands to gain

  • Herramientas de ingeniería de software – Los productos que generan código, como Claude Code, suelen detenerse durante minutos mientras el modelo procesa una solicitud. Reducir esa espera a unos pocos segundos haría que el desarrollo interactivo fuera mucho más fluido.
  • Flujos de diseño generativo – Los estudios que convierten prompts de texto en prototipos de aplicaciones o activos de juegos necesitan una iteración rápida para mantener el compromiso de los creadores. Una decodificación más rápida acorta los ciclos de diseño y aumenta las tasas de conversión.

Ambos sectores traducen la latencia directamente en pérdida de horas facturables o de clientes, por lo que un aumento de velocidad de 30× podría ser una ventaja competitiva decisiva.

El panorama general

La estrategia de Kog va en contra de la tendencia de la industria de construir silicio de IA personalizado. Empresas como Cerebras invierten miles de millones en chips que prometen un mayor rendimiento por vatio. Kog sostiene que las GPU actuales ya tienen suficiente ancho de banda de memoria para la decodificación; la pieza que falta es el software que realmente pueda utilizarlo. Si esta afirmación se mantiene a escala, los desarrolladores podrían posponer costosas actualizaciones de hardware y confiar en una actualización de software para lograr una inferencia casi en tiempo real.

Posibles obstáculos

  • Carga de mantenimiento – Cada nueva generación de GPU requerirá una nueva ingeniería inversa. A medida que el mercado se diversifica, el pequeño equipo podría verse sobrepasado.
  • Escalabilidad a modelos más grandes – La demostración utilizó un modelo de 2 mil millones de parámetros. Escalar las mismas técnicas a sistemas mucho más grandes puede chocar con límites de capacidad de memoria o exigir trucos de ingeniería adicionales que aún no se han revelado.
  • Caminos alternativos – Los proveedores de la nube ya ofrecen instancias optimizadas para la inferencia que combinan chips especializados y software. Para algunas cargas de trabajo, la ganancia marginal del stack de Kog puede no compensar la conveniencia de un servicio gestionado.

Qué observar

  • **Primer despliegue de un modelo