Deloitte Canada unveils the 2026 Technology Fast 50(TM) program winners as confidence and bold execution fuel growth. Oct 28, 2026, 05:00 ET For the 2026 Technology Fast 50 award category, Finni Health takes first place this year, achieving more than 46,000% revenue growth over four years. Across all 2026 winners, average revenue growth exceeded 3,000%. For the first time in the program's history, half of the companies in the top 10 Technology Fast 50 winners are led by women CEOs. TORONTO, Oct. 28, 2026 /CNW/ - Deloitte Canada is proud to announce the winners of its 2026 Technology Fast 50(TM) program. Now in its 29th year, the program recognizes the world-class achievements of Canada's fastest-growing technology companies and celebrates their confidence, resilience, entrepreneurial spirit and commitment to innovation. This year's Technology Fast 50 cohort achieved an average growth rate of 3,079%, surpassing the 3,000% mark for the first time in recent years. Despite tighter capital markets, economic uncertainty, and competitive pressure, the winners continue to invest, hire, innovate, and scale. Their achievements underscore the strength of Canada's technology sector and dispel any notion that tech firms struggle to achieve high growth while headquartered in Canada. "The strongest signal from the 2026 Technology Fast 50 class is confidence," says Brendan Cooper, Partner and National co-leader for the Technology Fast 50 program at Deloitte Canada. "These Canadian companies have been shaped by disruption and proven through crunch periods. That resilience is showing in how they lead today. They are investing, hiring, and innovating through uncertainty with confidence." Many of this year's fastest-growing companies were founded or began scaling around the pandemic, meaning their leaders have navigated disruption from the outset. That resilience is increasingly translating into bolder execution. In 2026, roughly 70% of applicants identified marketing and customer acquisition as a priority; more than 55% selected capital raising; and more than 50% identified strategic mergers and acquisitions as a priority. "This year's winners are moving from cautious ambition to bold execution," says Amanda Perran, Partner and National co-leader for the Technology Fast 50 program at Deloitte Canada. "These companies have excelled through their relentless focus on the fundamentals of growth: winning customers, attracting investment, building strong teams, and pursuing new opportunities. The next stage of their journey will be defined by how effectively they combine that ambition with the capital, talent, and AI-enabled capabilities needed to create sustainable, long-term value." Talent and innovation remain central to the growth agenda. Applicants identified workforce planning as their most pressing anticipated challenge, while technical upskilling and acquiring specialized skills emerged as leading talent priorities. At the same time, according to a new Deloitte report, close to 70% of qualifying companies devote a significant portion of revenue to research and development, and more than 85% expect AI to deliver productivity gains. 2026 winners: a new generation of Canadian technology leaders Finni Health tops this year's Technology Fast 50 category, posting four-year revenue growth of 46,574%. Headquartered in Toronto, Ontario, the company provides a platform that enables autism care providers (BCBAs) to launch, run, and scale their own private Applied Behavior Analysis (ABA) practices. In the Enterprise-Industry Leaders category, UniUni secured first place with four-year revenue growth of 2,143%. Headquartered in Vancouver, British Columbia, the company is a logistics company that provides delivery services to leading ecommerce and retail brands. Meanwhile, Turbopuffer was recognized as the number one winner in the Companies-to-Watch category, achieving three-year revenue growth of 65,443%, highest per cent growth in the National Capital Region in the past 5 years. Headquartered in Ottawa, Ontario, the company is a vector and full text search database purpose-built on object storage. These standout companies are not only advancing their respective industries but also reinforcing Canada's reputation as a source of globally competitive technology and innovation. Winner selection is based solely on revenue growth requirements. Each participating company must submit audited or reviewed financial statements to qualify for the program. Technology Fast 500(TM) ranking The Technology Fast 50 awards program runs alongside the broader North American Technology Fast 500(TM). This year, over 80 Canadian companies made the list. The complete list of Technology Fast 500 winners will be released on November 18. About the Deloitte Technology Fast 50 program The Deloitte Technology Fast 50 program is Canada's pre-eminent technology awards program. It recognizes business growth, innovation and entrepreneurship in three distinct categories: Technology Fast 50, Enterprise-Industry Leaders and Companies-to-Watch. The program also celebrates thriving technology companies in partnership with the North American Technology Fast 500(TM) program. Program sponsors for 2026 include RBCx, Norton Rose Fulbright, Export Development Canada, TMX, the Council of Canadian Innovators. For more information, visit www.fast50.ca. About Deloitte Canada At Deloitte, our Purpose is to make an impact that matters. We exist to inspire and help our people, organizations, communities, and countries to thrive. Our work underpins a prosperous society where people can find meaning and opportunity. It builds consumer and business confidence, empowers organizations to find imaginative ways of deploying capital, enables fair, trusted, and functioning social and economic institutions, and allows our friends, families, and communities to enjoy the quality of life that comes with a sustainable future. And as the largest Canadian-owned and operated professional services firm in our country, we are proud to work alongside our clients to make a positive impact for all Canadians. Deloitte provides industry-leading consulting, audit and assurance, tax, advisory and managed services to nearly 90% of the Fortune Global 500(R) and thousands of private companies. We bring together world-class capabilities, insights, and services to address clients' most complex business challenges. Deloitte LLP, an Ontario limited liability partnership, is the Canadian member firm of Deloitte Touche Tohmatsu Limited. Deloitte refers to one or more of Deloitte Touche Tohmatsu Limited, a UK private company limited by guarantee, and its network of member firms, each of which is a legally separate and independent entity. Please see www.deloitte.com/about for a detailed description of the legal structure of Deloitte Touche Tohmatsu Limited and its member firms. (C) 2026 Deloitte LLP and affiliated entities. SOURCE Deloitte Canada [email protected]

Fast, accurate retrieval with NVIDIA Nemotron 3 Embed. NVIDIA Nemotron 3 Embed models are now available in Baseten. Last updated. July 16, 2026 NVIDIA Nemotron 3 Embed helps AI systems find relevant information in large collections of text and code. The larger 8B model delivers leading retrieval accuracy, while the smaller, faster 1B model retains 95% of that accuracy. Both are now available on Baseten for AI agents, enterprise search, and code retrieval. Starting today, NVIDIA Nemotron 3 Embed 8B and NVIDIA Nemotron 3 Embed 1B are available on Baseten, giving developers two complementary options for balancing retrieval accuracy with indexing speed. Embeddings are the memory layer for retrieval. Retrieval starts by giving information an address. An embedding model converts data, like text, into a list of numbers called an embedding. In this mathematical space, passages with similar meanings sit close together, even when they use different words. A vector index stores these embeddings and makes them fast to search. When an agent or coding assistant needs more information, it can search the index to find the most relevant context for its task. Nemotron 3 Embed converts knowledge sources into embeddings stored in a vector index. It then embeds user queries to retrieve the most relevant context. Two common use cases show how this works in practice: Agentic retrieval: An AI agent can find relevant information across enterprise knowledge bases, documentation, and internal systems, helping it answer questions and complete tasks. Better retrieval leads to better context, and ultimately better answers. Code retrieval: A coding assistant can surface relevant files, functions, and dependencies across a large codebase, especially when the full repository can't fit in the model's context window. Better retrieval helps developers receive more accurate implementation guidance. Retrieval quality comes with a cost. Larger embedding models can produce higher-quality semantic representations, helping retrieval systems find more relevant information. However, they also require more processing time and hardware to generate embeddings. As documentation is revised, runbooks change, and pull requests land, slower indexing can leave a system searching an outdated version of its knowledge. Updates accumulate in an embedding queue. When the embedding model can't process changes quickly enough, the vector index becomes stale and the retrieval system may return outdated context. Agentic and code retrieval both face this tradeoff. Agents need to reason across sprawling enterprise knowledge, while coding assistants need to keep up with codebases that may change hundreds of times a day. Larger models can improve relevance, while smaller models can refresh the index faster as the underlying information changes. The best embedding model isn't simply the most accurate one. It's the one that balances retrieval quality with indexing speed for the workload. Two models for different priorities. NVIDIA designed Nemotron 3 Embed to combine high retrieval quality with the speed and efficiency needed for production workloads. Rather than forcing every use case into the same compromise, Nemotron 3 Embed offers two model sizes. Nemotron 3 Embed 8B delivers leading retrieval quality across major benchmarks - topping the RTEB leaderboard across both open and closed models. Choose it whenever retrieval quality is the highest priority. Nemotron 3 Embed 1B combines pruning - which removes less important parts of a model - and distillation - which trains a smaller model to reproduce the behavior of a larger one - to retain 95% of the 8B model's retrieval accuracy while substantially reducing serving cost and indexing latency. For NVIDIA Blackwell GPUs, the NVFP4 version delivers up to 2x higher throughput while retaining 99% of BF16 accuracy. Together, the two models let teams optimize for either maximum retrieval quality or production-scale efficiency. Choose based on what matters most. Choose Nemotron 3 Embed 8B when a missed or irrelevant result is mission critical. It's the better fit for complex enterprise collections of information and accuracy-critical workloads where maximizing retrieval quality matters more than how quickly the index can be updated. Choose Nemotron 3 Embed 1B when you need to index large amounts of content or refresh the index frequently. It generates embeddings faster and more efficiently while retaining 95% of the 8B model's retrieval accuracy. | Nemotron 3 Embed 8B | Nemotron 3 Embed 1B | | Leading retrieval accuracy | 95% of 8B retrieval accuracy | | Prioritizes finding the most relevant content | Prioritizes faster indexing | | Best for accuracy-critical workloads | Best for large or frequently changing content | Bringing Nemotron 3 Embed to turbopuffer. Choosing an embedding model is easier when teams can test it within their existing retrieval stack. That's why BaseTen Labs, Inc. is partnering with turbopuffer to make Nemotron 3 Embed available through its native embeddings. turbopuffer is a vector and full-text search engine built for AI applications. Its native embeddings reduce the engineering work required to deliver relevant semantic search results. That matters because getting search relevance right is hard. Teams need to compare models, tune their retrieval systems, and potentially re-embed documents as better options become available. By supporting existing and new, state-of-the-art embedding models, turbopuffer helps customers choose the right balance of retrieval quality, speed, and cost for their use case. As part of its partnership, BaseTen Labs, Inc. gave turbopuffer early access to Nemotron 3 Embed so the team could test the model and make it available through turbopuffer's native embeddings. BaseTen Labs, Inc. is excited to keep working together to help teams bring high-quality retrieval to production. Put Nemotron 3 Embed into production. Both Nemotron 3 Embed models are available in the Baseten Model Library for Dedicated Inference, giving teams a ready-to-deploy path to production without managing the underlying infrastructure. For workloads with specialized terminology, data, or relevance requirements, NVIDIA's Nemotron Embed fine-tuning recipe provides a path for adapting the models to a specific domain and can easily be run on Baseten Training. In-domain fine-tuning delivers ~10% accuracy improvement within just 5 hours. Fine-tuned models can then be packaged with Truss and deployed on Baseten. Whether you use a Model Library deployment or a fine-tuned model, once it's deployed, generate an API key and start creating embeddings for your knowledge sources.