December’s labour market numbers came in better than expectations, with the unemployment rate falling m/m from 6.9% to 6.7%. On face value, that’s good, right? Under the hood, the answer isn’t so clear.

Year-over-year, unemployment is still up big, sitting at the 2^nd highest level since September of 2021. The increase is broad-based, but key markets including Ontario and Quebec aren’t doing us any favours.

Perhaps more worrisome is the backdrop, which we think could be conducive to job loss that digs its heels in and is hard to move in the right direction.

Unemployment duration continues its climb, nearing 25 weeks on average and closing in on COVID-related highs from 2021. This time though, the share of long duration unemployment (>27 weeks) has grown gradually, indicating something structural.

Combine that with job vacancy rates continuing their march lower (now at ~3%), and you have a recipe for unemployment rates remaining higher for longer.

If the above link won’t work, try this: https://www.bullpen.finance/content/21

We got an update on the new housing price index on Friday, which showed prices softening 0.1% m/m, missing analyst estimates of 0.2% m/m growth.

For the most part we’re seeing a continuation of the trend as far as provincial impacts go, with growth everywhere but Ontario and B.C. as affordability concerns continue to drive interprovincial migration towards Alberta.

The headline numbers show just one data point, so it’s worth zooming out a bit to really capture what’s happening.

Some markets (Quebec, Alberta) continued to show strength post-stimulus, while others cooled off, weighing on the country average. This recent malaise in Canadian real estate appears here to stay for now, with a minimal reaction to rate cuts thus far.

The Canadian real estate market is at an interesting impasse, and we’ll make sure you don’t miss a thing.

Overnight futures are pointing downwards after Chinese company DeepSeek released R1, its frontier model that it claims is on par with OpenAI’s o1 model for a fraction of the cost.

NVIDIA (NVDA) finds itself in the eye of the metaphorical shit storm unfolding as markets re-assess their original CapEx assumptions linked to AI.

If DeepSeek is doing its final training runs for only ~$5.5M, how can companies like Meta and OpenAI justify investments in the hundreds of millions? And what happens to chip demand as training costs come down across the board?

The counter-argument is that if training costs fall by an order of magnitude, inference-related demand - the actual use of the models in the real world - will increase significantly because of how cheap they are.

This shift from training to inference could keep NVDA’s backlog healthy, but we’ll let the market decide. Prepare for volatility.