Together AI Marks Key Milestones at AI Native Event

Why It Matters

The performance and scalability breakthroughs lower cost and accelerate deployment of generative and agentic AI, giving customers a competitive edge. This solidifies Together AI’s strategic position in the rapidly expanding AI‑native startup ecosystem.

Key Takeaways

• 10x YoY revenue growth, 27 deals >$1M, one >$1B
• FlashAttention 4 gives up to 4x speed for long sequences
• New RL API separates inference from training for distributed pipelines
• ThunderAgent boosts agentic workload throughput 3.6x, reduces memory
• ATLAS-2 uses real-time data to deliver 1.5x faster inference

Pulse Analysis

The AI‑native cloud market is maturing as enterprises rush to embed generative models into core products. Analysts note that more than 70% of companies now use generative AI, making infrastructure performance a decisive factor for scaling. Together AI’s rapid 10× revenue growth reflects this demand, positioning the firm alongside other specialized cloud providers that focus on low‑latency inference and high‑throughput training for emerging startups.

Technical advances announced at AI Native Conf target the most pressing bottlenecks in modern model deployment. FlashAttention 4 narrows the gap between theoretical and real‑world efficiency, delivering up to four times faster processing for long‑context workloads such as code generation and document reasoning. The new Reinforcement Learning API decouples inference from training, enabling globally distributed pipelines without massive co‑located GPU farms, while ThunderAgent’s program‑aware scheduling pushes agentic workload throughput by 3.6× and slashes memory consumption. ATLAS‑2’s real‑time data adaptation further accelerates inference by 1.5×, illustrating how research can be rapidly productized.

Strategically, these capabilities give Together AI a defensible moat in a crowded cloud landscape. By bundling cutting‑edge research with production‑grade services, the company attracts high‑growth AI‑native firms that need both speed and reliability. Investors are likely to view the $1 billion contract and expanding customer base as validation of a sustainable business model, while competitors must now match the performance gains to stay relevant. As AI applications become more agentic and data‑intensive, infrastructure providers that can deliver measurable efficiency will dominate the next wave of AI‑driven innovation.