HubX at Google Cloud Day Türkiye

Cloud Day Türkiye brought together some of the most forward-looking teams in technology, and it was the perfect place for us to share the evolution of our work with Google Cloud. Our Co-Founder Kaan Ortabaş joined the stage to tell the story of how HubX shifted to a fully scalable AI infrastructure and how this transformation is helping us serve millions of users at the speed they expect.

The event also carried a special moment for our team. Google Cloud presented HubX with the first version of the TPU as a symbolic gesture of where this partnership is heading. It represents shared ambition, technical excellence, and a future that is being built together.

You can read the full Google Cloud case study here:

https://cloud.google.com/customers/hubx-ai

At HubX, we build apps that solve everyday challenges in simple and intuitive ways. From identifying plants to restoring photos or generating art, everything we create relies on AI that is fast and reliable.

Our apps like DaVinci, Momo, PlantApp, and PhotoApp reach millions of users in more than 190 countries. These users come with a clear expectation. They want instant answers. If the result arrives too late, the interaction breaks. This is one of the toughest realities of the mobile world. If an output takes 25 or 30 seconds to load, users do not wait. They move on.

This shaped a strict internal benchmark. Every AI workflow needs to stay below 10 seconds from the moment a user sends a prompt to the moment they see the result. This one metric influences engagement, retention, conversion, and the long-term health of every product we build.

Reaching that threshold consistently required a deeper and more scalable infrastructure. That is where our partnership with Google Cloud began.

Before GKE, our infrastructure struggled under heavy AI workloads. As search volumes increased and global traffic grew, we faced delays, higher churn, and slower iteration cycles. We needed a platform that could scale instantly, support large AI models, and stay cost-efficient without slowing down development.

Google Kubernetes Engine created a foundation that aligned with everything we needed. The scalability, reliability, and flexibility of GKE allowed us to rebuild our systems in a way that matched our pace of innovation.

The transition was smooth and fast. Google for Startups Cloud Program supported us with technical training, hands-on guidance, and cloud credits. This eliminated friction and helped us migrate workloads directly into GKE.

The collaboration immediately delivered measurable impact across speed, operations, and user experience.

• 2.5x faster inference speeds using Trillium TPUs and GKE

• 40 per cent lower operational costs by optimising AI and ML workloads

• Less than 10 seconds from query to response time across major app flows

• 20 to 30 times faster model deployment time

• Significant improvements in cold start times, reduced by an entire minute

• Smooth scalability during peak traffic generated by trends or marketing campaigns

This meant that every product team began working faster. Every experiment reached production quicker. Every update became easier to ship. And every user saw the impact directly in their experience.

Our technical stack today reflects the flexibility we gained from GKE and AI Hypercomputer.

We use a combination of accelerators to balance power and cost.• TPUs for fine-tuning and inference in highly demanding AI workflows• A100 GPUs for large, complex model processes• L4 GPUs for lighter, cost-optimised workloads

This mix allows us to match resources with the exact needs of each model. The shift between TPU and GPU environments happens seamlessly inside GKE, giving our developers freedom to scale without technical overhead.

Cloud Run plays a major role in fast prototyping. Teams deploy new features in a serverless environment, test performance instantly, and migrate the successful ones into GKE when they are ready for long-term scaling.

Hyperdisk ML improved storage performance. Large model images now load far faster, even during heavy traffic. Boot-up times improved by 20 to 30 times, which keeps our infrastructure stable during rapid spikes in demand driven by trends or global app moments.

The mobile ecosystem changes every day. Search trends shift, new categories rise suddenly, and global interest can multiply within hours. Our infrastructure now moves at the same speed.

With GKE, our team can

• Respond to market shifts faster

• Launch new features with less risk

• Handle heavy AI workloads without latency

• Keep costs predictable and optimised

• Ensure consistency for millions of global users

This is especially important during performance marketing campaigns where traffic scales at a pace that only a fully automated system can handle.

Our partnership with Google Cloud is a long-term commitment to speed, intelligence, and global scale. Everything we build at HubX carries the same ambition. Create products that people love. Build technology that stays ahead of expectations. And design experiences that feel instant, simple, and accessible to everyone.

With GKE and Google Cloud, we are pushing that ambition even further. Our infrastructure now reflects the scale of our vision and the momentum of our team. Onward to faster, smarter, and more scalable apps.