Canadian AI: Taalas and LawZero raising $100M+ each
And Canadian AI in labs, corn fields and mines.
Good morning! Welcome to the Canadian AI Newsletter, a weekly roundup for founders, operators and investors.
Was this forwarded to you? Subscribe below.
The Canadian AI Newsletter is written by Raif Barbaros, Partner at Mistral Venture Partners. Views are my own.
This was the week a Toronto chip startup nobody had heard of raised more than most Series B rounds. Taalas pulled in $169M to hardwire AI models directly into silicon — the kind of deep tech bet Canada rarely makes at this scale. Cohere shipped a multilingual model that runs on your phone in 70+ languages. Robots showed up too, in Montréal labs, on Ontario corn fields, and underground in Sudbury mines. Let’s get into it.
💰 Deals & Milestones
Taalas (Toronto) raised $169M to build model-specific AI chips that hardwire inference directly into silicon.
Founded by Ljubisa Bajic (Tenstorrent co-founder). Team of 25 engineers from AMD, Apple, Google, Nvidia.
First chip (HC1) runs Llama 3.1 8B at 17,000 tokens/sec — 73x faster than Nvidia’s H200 at 1/10th the power.
Partners with TSMC on a two-month chip turnaround vs. ~six months for a standard AI processor. Plans to handle frontier models by year-end.
Backers include Quiet Capital, Fidelity, and chip industry VC Pierre Lamond. Total raised now $219M.
LawZero (Montréal) — Ottawa signed a letter of intent to invest more than $100M in Yoshua Bengio’s AI safety non-profit.
LawZero is building actual AI models and technical solutions for safe, trustworthy AI. Bengio: “This is not a university project, but something that’s pushing the frontier of AI.”
Barely a year old, ~30 employees, plans to top 100 next year. Had launched with US$30M in funding. Most of the new capital goes to compute.
Minister Solomon: “This is a bet we want to make. We want to support Canadian tech.” One of the largest federal AI investments since the $240M to Cohere.
Haply Robotics (Montréal) raised $16M CAD to build haptic control systems for physical AI.
150+ customers, including 30 Fortune 500 companies. Revenue growing 150% YoY. Won CES 2026 awards.
CEO Colin Gallacher: “There are going to be 10 billion robots on the planet by 2050. They’re going to need steering wheels.”
Led by Sound Media Ventures. Amazon Industrial Innovation Fund, Hanwha, Two Small Fish Ventures, BDC Capital Deep Tech, all in.
Shakudo (Toronto) raised $7M USD for its AI deployment platform for regulated industries.
Series A2 Round led by Wittington Ventures, family office of the Weston family, owners of Loblaw and Shoppers Drug Mart. Loblaw is a customer. Shakudo had raised a $7M in a Series-A in 2023.
Upside Robotics (Waterloo) raised $7.5M seed for autonomous farming robots that deliver precision fertilizer to crops.
10,000+ autonomous km logged, 100,000+ litres of fertilizer applied. Cuts fertilizer use by 70%, saving farmers ~$150/acre/season. 200 farms on the waitlist.
Led by Plural with Garage Capital and the founders of Clearpath Robotics.
Toyo raised $4.3M seed for an agent-native platform that replaces SaaS tools with AI agents for founders.
Led by Frontline Ventures with iNovia Capital, Tiny Supercomputer, and angels from Amazon, Microsoft, Cloudflare.
Serial founders. Previous exits include Pusher (to MessageBird) and MediaCore (to Workday).
LoopX (Sudbury) received a $480K FedNor grant to commercialize generative AI analytics for mining.
Real-time decision-making and operational awareness underground. AI in the resource economy, the kind of story nobody else covers.
Cohere (Toronto) launched Tiny Aya, a family of open-weight multilingual models supporting 70+ languages.
3.35B parameters. Runs on a laptop offline. Regional variants for Africa, South Asia, Asia-Pacific. Launched at the India AI Summit — a strategic play for the next billion users outside the English-speaking world.
(Disclosure: Cohere is a portfolio company.)
RBC created a new AI Group reporting directly to the CEO, targeting $1B in enterprise value from AI by 2027. Given their market cap of $300B+, not the most ambitious of aspirations, but whatevs.
Also partnering with Cohere on North for Banking, foundational enterprise AI models for financial services.
Shopify launched its Winter ‘26 Edition, its twice-yearly product release: it was about agents, agents, and, also, agents.
Products now surface directly inside AI conversations on ChatGPT, Perplexity, and Microsoft Copilot. One setup, every agent.
Tobi Lütke: “We’re making every Shopify store agent-ready by default.”
Ben Thompson called Shopify “one of the biggest winners from AI.”
🏛️ Policy
OpenAI and the Tumbler Ridge crisis. Shooter was an OpenAI user. OpenAI closed his account due to misuse related to violence planning. But OpenAI never came to the Canadian authorities until after the shooting. Ottawa summons OpenAI.
Minister Evan Solomon’s three-continent diplomatic tour. Solomon hit Munich, New Delhi, and Washington in a single week — signing the Canada-Germany Sovereign Technology Alliance, attending the India AI Summit, and meeting US counterparts. The most active week of AI diplomacy Canada has had in years.
Bank of Canada’s Carolyn Rogers told businesses to take a leap of faith on AI at the Rotman Productivity Conference — invest now or get left behind.
BMO chief economist Douglas Porter responded by calling AI productivity gains overhyped, arguing that Canada’s productivity crisis is the norm, not new, and that the internet didn’t deliver the boosts people assumed it would either.
So the gov’t rep is urging businesses to be aggressive and take a leap of faith, while the private-sector economist is telling them to be conservative. 🤦♂️
🔬 Research
Geoffrey Hinton gave a long-form interview to CBC Ideas (Feb 18) arguing AI needs to develop “maternal instincts” — caring for humans the way a mother cares for a child — rather than being designed as a submissive assistant. His framing: “If it’s not going to parent me, it’s going to replace me.”
Yoshua Bengio told Business Standard at the India AI Summit, “Forget AGI if AI models stay inconsistent with jagged skills.” The definition of AGI and the degree of jaggedness it can exhibit is really at the heart of this question. For what it’s worth, humans are pretty jagged in their intelligence, too.
Vector Institute hosted its third annual Remarkable conference (Feb 19-20) featuring Cohere co-founder Nick Frosst as a Day 1 speaker, alongside researchers from Google DeepMind, Thomson Reuters, and the Acceleration Consortium. Also released its MIDST privacy paper — a practical framework for deploying differential privacy in production ML pipelines.
📊 Data
TELUS cross-border AI report: 85% of Canadians are using AI. 90% believe AI should be regulated. Similar numbers in the US. 11,000+ surveyed.
Subscribe for weekly delivery:
Have a great week! See ya 👋