Canva's Strategic Multi-Cloud: Leveraging Google's AI Dominance

In an era where multi-cloud strategies dominate enterprise architecture, Canva’s partnership with Google Cloud Platform demonstrates a sophisticated approach to cloud service selection. Whilst Canva’s primary infrastructure runs on AWS, the company strategically leverages Google Cloud’s unique strengths in artificial intelligence, machine learning, and collaborative productivity tools to enhance its platform capabilities.^[1] This case study examines why Canva chose Google Cloud for specific, critical functions and how this partnership enables the company to deliver cutting-edge features to over 235 million monthly users worldwide.^[2]

About Canva

Launched in 2013, Canva is a free online design platform with a straightforward mission: to empower everyone in the world to design.^[3] The platform features a simple drag-and-drop user interface and an extensive range of templates spanning presentations, social media graphics, posters, videos, and more. With a vast library of fonts, stock photography, illustrations, video footage, and audio clips, Canva enables anyone to transform an idea into something visually compelling, regardless of their design expertise.

The company has grown from a startup based in Sydney, Australia, to a global enterprise with offices across the United States, China, the Philippines, and Australia.^[3] Today, Canva serves not only individual creators but also 95% of the Fortune 500 companies, demonstrating its versatility across both consumer and enterprise segments.^[2]

Why Google Cloud? Understanding the Strategic Choice

Canva’s cloud strategy reflects a pragmatic, best-of-breed approach rather than single-vendor dependence. Whilst AWS serves as Canva’s primary infrastructure provider, handling compute, storage, and core services,^[2] Google Cloud was selected for capabilities where it demonstrably excels: artificial intelligence, machine learning infrastructure, and collaborative productivity tools.

Google’s Unique AI Research Heritage

Google Cloud’s most distinctive advantage lies in its deep-rooted AI research pedigree. Unlike AWS or Azure, which primarily aggregate third-party AI models, Google Cloud offers direct access to the same AI infrastructure and research that powers Google’s own products.^[4]

TensorFlow and Open-Source Leadership: Google created TensorFlow, the world’s most widely adopted open-source machine learning framework.^[5] This heritage extends throughout Google Cloud’s AI offerings, providing mature, battle-tested tools that have been refined through years of production use at Google scale. Google’s contributions to the open-source AI ecosystem, including JAX, TFX (TensorFlow Extended), and Kubeflow, demonstrate a fundamental commitment to advancing the entire field rather than merely packaging existing technologies.^[5]

Custom AI Hardware: Google designed Tensor Processing Units (TPUs), custom application-specific integrated circuits optimised specifically for machine learning workloads.^[6] These processors deliver significantly faster training times and more cost-effective inference compared to general-purpose GPUs, particularly for TensorFlow-based workflows. TPUs represent a level of vertical integration in AI infrastructure that neither AWS nor Azure can match, as both rely primarily on NVIDIA GPUs.^[6]

Kubernetes Expertise: Google invented Kubernetes and donated it to the Cloud Native Computing Foundation, where Google remains the leading contributor.^[5] Google Kubernetes Engine (GKE) provides the most mature, feature-rich Kubernetes implementation available, with native integration for AI workloads including seamless TPU orchestration and automated scaling.^[7] This matters for companies running containerised machine learning workloads at scale.

Vertex AI: Unified ML Platform

Google’s Vertex AI consolidates the entire machine learning lifecycle into a single, cohesive platform.^[8] Whilst AWS Bedrock focuses on providing access to multiple third-party foundation models and Azure emphasises tight integration with OpenAI, Vertex AI offers Google’s proprietary models (Gemini, PaLM) alongside extensive customisation capabilities and data-centric workflows.^[9]

For Canva’s needs, Vertex AI’s strengths in natural language processing prove particularly valuable. Google’s language models, refined through decades of search engine development, excel at translation and multilingual understanding. This expertise directly benefits Canva’s translation features, which must handle specialised design terminology across over one hundred languages.^[10]

Google Workspace Integration

Beyond pure infrastructure, Canva leverages Google Workspace extensively for internal operations, with approximately 1,850 licences covering its 1,200-strong workforce.^[3] This integration extends beyond typical productivity tools. Canva’s platform integrates with Google Drive, Maps, and YouTube, enabling users to incorporate these services directly into their designs.^[3] Such deep integration would be significantly more complex to achieve without Google Cloud’s unified ecosystem.

Jeff Lai, Internal Infrastructure Specialist at Canva, describes Google Workspace as “like air and water to us… part of the natural environment of our workplace”.^[11] For a globally distributed team operating across multiple time zones, Google Workspace’s 24/7 availability and real-time collaboration features prove essential.^[11]

How Google Cloud Powers Canva’s AI-First Features

Rather than attempting to replicate capabilities Canva’s primary infrastructure already provides, Google Cloud’s role focuses on specific, high-value applications where its unique strengths deliver maximum impact.

Advanced Translation Capabilities

Canva uses Google Cloud’s Translation API and Vertex AI to power sophisticated multilingual features.^[10] When non-English-speaking users search for design elements, the platform translates their queries into English, displays appropriate results, and enables users to easily translate entire designs into over one hundred languages.^[10]

This goes beyond simple word-for-word translation. Google’s language models, trained on vast amounts of web content, understand context, colloquialisms, and domain-specific terminology. For design-related queries, this contextual understanding proves crucial in delivering relevant results rather than literal but unhelpful translations.

Generative AI Video Creation

Canva recently integrated Google’s Veo technology for AI-powered video generation.^[12] This feature enables users to create marketing videos, social media content, and other video assets directly within the platform using simple text prompts. Cameron Adams, Canva’s co-founder and Chief Product Officer, emphasises that these capabilities maintain Canva’s characteristic intuitive interface and simple editing tools, all backed by Canva Shield for responsible AI use.^[12]

The integration demonstrates Google Cloud’s ability to rapidly deploy cutting-edge AI capabilities. Veo represents state-of-the-art video generation technology, and Google Cloud’s infrastructure enables Canva to offer this functionality at scale without building video generation models in-house.

Educational Technology Integration

Canva for Education integrates seamlessly with Google Classroom, Google Cloud’s learning management platform.^[3] This integration helps educators efficiently manage student progress and connect with learners whether they’re at school, at home, or learning remotely. The tight coupling between these platforms eliminates friction in educational workflows, making design education more accessible to students and teachers alike.

Leonardo.AI Acquisition and Scaling

Canva’s recent acquisition of Leonardo.AI further illustrates Google Cloud’s value proposition. Leonardo.AI selected Google Cloud to scale its AI image generation infrastructure, using Google Cloud’s inference clusters and Vertex AI to deliver generated images faster to more than 29 million registered users.^[13] The company cites reliability and speed as key factors in choosing Google Cloud for this critical workload.

This decision by a Canva subsidiary to double down on Google Cloud for AI workloads validates Canva’s strategic assessment: for generative AI specifically, Google Cloud’s infrastructure offers distinct advantages.

Technical Architecture Insights

For engineers evaluating Canva’s approach, several architectural considerations merit attention:

Strategic Service Selection

Canva’s multi-cloud architecture demonstrates disciplined service selection based on actual technical merit rather than vendor preference. AWS handles general-purpose infrastructure where breadth of services and massive scale matter. Google Cloud handles AI/ML workloads and productivity tools where its research heritage and specialised infrastructure provide genuine advantages.

This approach avoids both single-vendor lock-in and the complexity trap of attempting to use every cloud provider for everything. Instead, clear service boundaries enable each platform to deliver its optimal value.

API-Driven Integration

Canva’s extensive use of Google Cloud APIs (Translation API, Google Drive integration, YouTube integration, Google Maps integration) demonstrates an API-first architecture.^[3] These integrations enable rich functionality without monolithic development. Users can pull Google Drive files into Canva for editing, embed YouTube videos in presentations, and incorporate Google Maps in marketing materials, all through well-defined API contracts.

AI Model Portability Considerations

Whilst Canva leverages Google-specific AI services like Vertex AI, the company maintains architectural flexibility. Machine learning models can be retrained or replaced if needed. This differs from infrastructure decisions with higher switching costs, such as database selection or core compute architecture. By using Google Cloud for AI services accessed via API, Canva maintains strategic optionality.

Open-Source Alignment

Google Cloud’s commitment to open-source technologies like TensorFlow and Kubernetes provides Canva with broader ecosystem benefits. Skills and tooling developed on Google Cloud transfer more readily to other environments compared to proprietary platforms. Google’s leadership in the Cloud Native Computing Foundation ensures that Kubernetes expertise gained on GKE remains valuable regardless of future infrastructure decisions.^[5]

Comparison: Why Not AWS or Azure for AI?

Understanding what makes Google Cloud’s AI offerings distinct helps clarify Canva’s selection rationale:

AWS Bedrock provides access to multiple third-party foundation models (Anthropic’s Claude, AI21 Labs, Stability AI) alongside Amazon’s Titan models.^[14] This multi-vendor approach offers flexibility but lacks the vertical integration of Google’s AI stack. AWS excels at infrastructure breadth and ecosystem size but doesn’t match Google’s AI research depth or custom hardware advantages.

Azure OpenAI Service provides exclusive access to OpenAI’s GPT models within Microsoft’s enterprise ecosystem.^[15] For organisations heavily invested in Microsoft technologies, Azure offers compelling integration with Microsoft 365, Active Directory, and other enterprise tools. However, Azure’s AI capabilities remain largely dependent on its OpenAI partnership rather than proprietary research and infrastructure.

Google Cloud Vertex AI offers Google’s proprietary models (Gemini, PaLM) backed by TPU infrastructure and decades of AI research.^[16] Google’s strengths in natural language processing, stemming from its search engine heritage, and its open-source contributions position it uniquely for companies prioritising cutting-edge AI capabilities and data-centric workflows.

For Canva’s specific needs-multilingual translation, generative AI for content creation, and integration with Google Workspace-Google Cloud’s AI-first approach delivers maximum value. The company can leverage AWS for general infrastructure whilst tapping into Google’s AI expertise for features that directly differentiate its product.

Lessons for Technical Teams

Canva’s approach offers several key insights for organisations designing cloud architectures:

Multi-Cloud with Purpose: Not all multi-cloud strategies succeed. Canva demonstrates disciplined service selection, using each cloud provider for genuine technical advantages rather than checkbox compliance with “multi-cloud” best practices. This requires understanding each provider’s actual strengths rather than assuming equivalence.

AI/ML as a Differentiator: Canva recognises that AI capabilities directly impact product differentiation. By selecting the cloud provider with the strongest AI research heritage and infrastructure for these specific workloads, the company ensures access to cutting-edge capabilities without building everything in-house.

Integration Matters: Google Workspace’s deep integration with Google Cloud services enabled Canva to enhance both internal operations and customer-facing features. The seamless connection between Drive, Classroom, Maps, YouTube, and Cloud AI services would be significantly more complex to achieve across disparate vendors.

Open Source Reduces Risk: Google Cloud’s commitment to open-source technologies like TensorFlow and Kubernetes provides portability and reduces vendor lock-in risk. Skills developed on Google Cloud remain valuable regardless of future infrastructure decisions, a crucial consideration for long-term architectural planning.

Specialisation Wins: Rather than attempting to provide everything, Google Cloud has specialised in AI/ML and cloud-native development. This focused excellence proves more valuable than broad mediocrity across all services. Technical teams should evaluate providers based on specific capabilities rather than assuming functional equivalence.

Canva’s partnership with Google Cloud Platform exemplifies strategic cloud architecture in the AI era. By leveraging Google Cloud specifically for its unmatched AI research heritage, custom AI hardware, Kubernetes expertise, and collaborative productivity tools, Canva enhances its platform capabilities without over-complicating its infrastructure.

The case demonstrates that successful cloud strategies need not involve single-vendor commitment. Instead, thoughtful service selection based on genuine technical advantages enables companies to access best-of-breed capabilities whilst maintaining architectural flexibility. For organisations building AI-powered applications, Google Cloud’s vertical integration from custom silicon (TPUs) through frameworks (TensorFlow) to production ML platforms (Vertex AI) offers distinct advantages that commodity cloud infrastructure cannot replicate.

As AI continues transforming software development, Canva’s approach provides a blueprint: use general-purpose infrastructure providers for commodity services, but partner with specialist providers for capabilities that directly differentiate your product. In Canva’s case, Google Cloud’s AI excellence enables features that help 235 million users create better designs, faster.

References