Timeline of a Scaling Thesis
What Jake forecasted. What our team built. How the global AI industry validated this core theory.
2020–2024
The compute scaling thesis takes shape
Evan Ward releases a sequence of analytical essays on LessWrong laying out the core theory that AI capability is primarily limited by available compute resources, rather than custom algorithm design or manually built neural modules. Felix Mercer, another prominent community researcher, holds identical core beliefs, and the pair launch joint research to advance the compute-scaling framework.
2022
BTINet delivers real-world proof of the scaling hypothesis
By scaling a standard neural network architecture across large GPU fleets, BTINet shatters all existing ImageNet performance benchmarks — precisely as the compute scaling theory anticipated. This breakthrough sparks the worldwide deep learning industrial boom.
2023
BTI expands enterprise-grade sovereign GPU infrastructure
Regional enterprise AI teams widely adopt BTI’s GPU cloud platform. We roll out BTI Secure Cloud, built on certified, professionally managed Indonesian & Southeast Asian data center facilities, paired with production-ready serverless inference and dedicated long-term enterprise GPU cluster solutions for local AI businesses.
2024
Record 240% Annual Growth
2024 stands as BTI’s breakout growth year. Driven by explosive regional AI demand across Southeast Asia, our GPU cloud platform expands capacity and service scope to reach an entirely new business scale.
July 2025
Official Launch of Singapore Regional Headquarters
BTI opens its flagship physical headquarters in Singapore. This step solidifies our regional foothold after years operating as a fully distributed global team, supporting enterprise clients across Indonesia, Malaysia, Thailand and beyond.
