Priya Raman
Verified WriterSenior AI & Tech Correspondent
Covers applied AI, GPU economics, and how foundation-model shifts reshape North American product roadmaps.
Biography
Priya Raman covers the applied AI beat — model releases, inference economics, GPU supply, and the operational reality of shipping AI features inside North American startups and incumbents. She maintains an open methodology document explaining how she benchmarks model quality, latency, and cost.
Before Vanhub, Priya led a small ML platform team at a B2B SaaS unicorn, owning the migration from in-house GPU clusters to managed inference. She brings hands-on engineering context to coverage that is otherwise dominated by hype.
Priya holds an M.S. in Computer Science (ML systems) from the University of Toronto and contributes occasional code to open-source LLM tooling.
Latest stories by Priya Raman
API Lock-In at $30K/Month: Vancouver Firms Bet on Open-Weight AI
With Bill C-27 dead and CLOUD Act exposure real, Vancouver mid-market firms face a 2026 inflection point on AI infrastructure — and waiting costs more than deciding.
21%: BC Universities Lost the GPU Lottery — and Now They're Paying AWS
Canada's sovereign AI compute strategy sent $42.5M to U of T first. BC researchers got 21% of requested GPUs and a 7.5% brokerage fee. Here's who's actually paying.
65,372 Exits — and Now a $100K U.S. Fee Sending Engineers Home
Canada's brain drain hit a 50-year record in 2024-25. A six-figure H-1B fee and Ottawa's $1.7B talent strategy are opening a narrow window — Vancouver has 18 months to capture it.
Ottawa Spent $2B on Sovereign AI Compute. Azure Is Still Winning.
Federal and BC dollars are flowing to SFU and UVic supercomputers — but the fine print of Canada's compute strategy is actively subsidizing the hyperscaler dependency it claims to fix.
$300M to AWS: Canada's Sovereign AI Compute Plan Has a Contradiction
Ottawa committed $2.9B to build Canadian-owned AI infrastructure — then wrote a nine-figure cheque to US hyperscalers. BC universities are caught in the gap.
Cheap Tokens, Rising Bills: The Break-Even Trap Killing AI Startups
Inference costs fell 1,000× in three years — yet total AI bills keep climbing. Here's the math Vancouver founders aren't running before they burn through seed funding.
41% Margins, Zero Dilution: The Bootstrapped AI SaaS Playbook
Canada's $300M compute subsidy quietly eliminated bootstrapping's biggest cost barrier. Here's the math Vancouver AI founders are running right now.
API Costs in 2026: The 4x Price Gap That's Actually a Decoy
OpenAI, Anthropic, and Google's token rates diverge by 4x — but Vancouver B2B startups optimizing list price are solving the wrong problem entirely.
42% of AI Projects Failed in 2024. Canada's Make-vs-Buy Bet Is Next
Canadian enterprises face a strategy crisis disguised as a tech decision. Here's what the abandonment rate, consumption pricing, and Bill 31 actually mean for your AI budget.
4% Success Rate, $2.8M Bill: The AI TCO Trap Mid-Market Missed
Microsoft Copilot seats surged 160% but only 4% of deployments generate real value. Here's the full cost math Canadian mid-market CFOs aren't running.