Stephen Bacso

A Canadian startup just printed an AI model into a chip. Physically etched into silicon. 53 billion transistors. Been following Taalas for a while and their HC1 chip hits wild numbers: 17,000 tokens/sec per user. That's ~74x faster than Nvidia's H200. More than 8 times faster than the current state of the art. $0.0075 per million tokens. That's 2,000x cheaper than cloud APIs. 20x less the cost to build, and 10x lower power consumption. But here's the best part. Despite the model being physically hardwired into the chip, it supports LoRA (Low-Rank Adaptation) fine-tuning. One chip, many specialisations. Load a legal LoRA. It runs legal AI. Swap in a medical one. It runs clinical AI. Same silicon with the same embedded model, completely different behaviour. This is exactly what we've been building toward with Ertas AI. Fine-tuning isn't just a training technique anymore. It's becoming a hardware deployment interface. The adapter you create today runs on GPUs now and on dedicated silicon tomorrow. The teams learning to fine-tune now will have production-ready models when this hardware ships at scale. Everyone else will be starting from scratch. P.S. Wrote a full analysis with benchmarks and comparison charts on our blog. Link in comments. #AI #hardware #inference #finetuning

685

41 Comments

The Canadian Quantum Champions program is funding startups like Anyon Systems Inc., Xanadu, Photonic Inc., and Nord Quantique. Is this a bold push to avoid repeating our AI missteps, or just the beginning? John Ruffolo and I discuss this and the rest of the headlines announced in the last few days on Tank Talks, including: ✅ Microsoft's massive $7.5B investment in Canadian AI and cloud infrastructure—sparking tough questions on true data sovereignty under the U.S. Cloud Act ✅ We unpack layoffs hitting consulting giants like McKinsey as AI commoditizes knowledge work, shifting value to trust, execution, and new models like forward-deployed engineers. ✅ Plus, crypto's wild play: Tether eyeing a stake in Juventus Football Club, raising eyebrows on regulation and asset backing. Don't miss this insightful breakdown on building sovereign tech, the future of work, and trust in the AI era. Watch and listen now [links in comments] #TankTalks #CanadianInnovation #AI #QuantumComputing #VentureCapital #RippleVentures

30

5 Comments

One of Canada's most underappreciated #innovation stories right now may be how broadly we're competing in the frontier computing arena, and how differently that looks across our ecosystem. Some Canadian companies are building the frontier itself: 👉 Photonic Inc. Inc. — distributed quantum computers on silicon spin qubits. 👉 Nord Quantique — superconducting quantum hardware with novel error correction. 👉 Xanadu — pioneering photonic quantum computers and chips. 👉 D-Wave — Canadian-founded, the only company pursuing both annealing and gate-model quantum hardware simultaneously. 👉 Cohere — redefining enterprise AI at scale. But there's a second Canadian story that warrants attention — companies harnessing frontier computing R&D to strengthen their moat in their own domain, using novel, world-class algorithms as a distinct and deliberate innovation and productivity advantage. The most compelling recent example? OTI Lumionics Years ago, I had the pleasure of meeting Scott Genin and the OTI team through my work at Fujitsu, exploring the frontiers of quantum-inspired computing. OTI is fundamentally an advanced materials company, developing next-generation OLED materials for consumer electronics and automotive applications. But to get there faster and better than anyone else, they built their own world-class computational engine. Their latest result is a home-grown computational breakthrough with significant implications for chemistry, materials science, and perhaps the future of quantum computing itself. Canada is building something real here. At every layer of the stack. Scott's post and paper are in the comments. If you're in the arena, well worth your time. 🇨🇦 #CanadianInnovation #DeepTech #FrontierComputing #QuantumComputing #MaterialsScience #OLED

16

6 Comments

Vention , the leading digital-first industrial automation platform, announced the opening of its Physical AI lab in Montreal. The lab is led by Jimmy Li, Vention's Director of Physical AI, and advised by Joelle Pineau Chief AI Officer Cohere, and focuses on robotic manipulation for manufacturing, including industrial goods, electronics, and automotive production. https://lnkd.in/gmCEcpv5

55

I spoke with Stojancho of Aurora Labs here at ROSCon 2026 here in Toronto. Aurora Labs is presenting LOCI: an AI-driven execution intelligence and performance observability platform designed to analyze compiled binary code and predict software behavior. These videos are sponsored by MHI: The Association That Makes Supply Chains Work , The Founders Guild, Intelligent Machine Guild, Mekable, Goldman Sachs, General Motors, Snell & Wilmer, Pillsbury Law Firm, Optimist Legal, Hertz Foundation, Massachusetts Institute of Technology, Stanford Robotics Center, City of Half Moon Bay, and Draper University Thank you to Vanessa Yamzon Orsi, Geoffrey Biggs, Tully Foote, Katherine Scott, Ryan Gariepy, Brian Gerkey, Open Robotics, and the whole ROSCon organizing team for putting this together in Toronto. 
To see more great automation companies follow Daniel Theobald, and subscribe by pressing the bell on my profile. Please comment, like, and repost these videos to get the word out about these great companies. Starting something? Join 10,000+ founders at The Founders Guild: foundersguild.org Building intelligent machines? Join the work at the Intelligent Machine Guild: https://lnkd.in/g62-CPWM #ROSCon #FoundersGuild #Robotics #Automation #IntelligentMachineGuild #AI #Mekable #IntelligentMachineGuild


2