1. US manufacturing sector improved in November.

Though the absolute level of the index still remains below 50 and points to ongoing weakness/contraction in manufacturing.

Monday's reading could be an early sign that "better days lie ahead," according to Jefferies US economist Thomas Simons.

2. The CAC 40 is currently set for its worst annual underperformance to Germany’s DAX.

If Michel Barnier’s government falls, not only will the adoption of the 2025 budget be delayed, but the debate in France will switch to the next presidential election.

3. The “fetishization” of US Stocks.

Ownership of US equities across households, financials and foreigners is at all-time highs, while valuations have hit a record peak.

Never before has ownership been so high with valuations so stretched.

Equity investors are expecting rates to fall AND stocks to rise AND valuations to keep climbing.

Historically speaking, this could turn out to be one of the worst times to invest for the longer term.

4. The market is more vulnerable to a decline.

Long-term average returns in the S&P are set to worsen in the coming years as the almost inevitable consequence of the obsession with US stocks.

Aggregating across sectors, the exposure to equities as a proportion of financial assets is close to a record high. (Blue line on a reversed left scale)

When everybody’s long, there are fewer people who need to buy - or are willing to do so at higher prices.

This is what has happened in the past.

5. Here are Morgan Stanley's forecasts for the G10 currencies in 2025.

6. “The real commercial value comes with inference, and inference is starting to gain scale,” says Cristiano Amon, the chief executive of Qualcomm.

Amazon, Advanced Micro Devices and several start-ups are beginning to offer credible alternatives to Nvidia’s chips, especially for a phase of A.I. development known as inferencing.

In the field of artificial intelligence (AI), inference is the process that a trained machine learning model uses to draw conclusions from brand new data . An AI model that can make inferences can do so without any examples of the desired outcome.

That process, called “inferencing,” happens after companies use chips to train A.I. models.

It allows them to carry out tasks such as serving up answers with A.I. chatbots.

For these inferencing tasks, Nvidia’s rivals are proving they can deliver much faster speed, and at prices that are much lower, said Daniel Newman, an analyst at Futurum Group.

Not a subscriber yet?