Sat, 07 Feb 2026 04:35:00 +0000
Enterprise SaaS spending per employee will fall from roughly 10% of an employee's salary toward 5% and eventually toward 1%, forcing SaaS companies to significantly downsize their expense base.
“So your SAS spend might go from 10% of an employee's salary down to 5% down to 1%. That's what I think the trend will be which means these companies are going to need to really downsize their expense base.” View on YouTube
A gradual, multi-year trend with no clear endpoint reached yet.
Within 5 years (by roughly 2031), the combined market capitalization of today's software companies will be 4x to 10x higher than current levels, though gains will be unevenly distributed across companies.
“I would say that you could probably take the sum of the market cap of all the software companies today and have a pretty good bet that everything will be four to 10x higher 5 years from now, but it's going to be not evenly distributed.” View on YouTube
The predicted five-year software market cap horizon has not yet elapsed.
In 2026, incoming Fed Chair Kevin Worsh will deliver more interest rate cuts than the market currently expects.
“I happen to think that I would take the over on the number of rate cuts that Worsh is going to give us this year.” View on YouTube
Incoming Fed Chair Kevin Warsh confirmed a hawkish policy stance and signaled hikes rather than cutting more than the market expected.
Over the next 6 months to a year (roughly through early-to-mid 2027), Kevin Worsh as Fed Chair will push to cut interest rates.
“I do think that Worsh has been consistent for the last year, saying the Fed was taking too long to realize that inflation is falling and that they should be cutting more. So, I do think that over the next, say, 6 months to a year, he's going to want to cut rates.” View on YouTube
Kevin Warsh confirmed a hawkish stance as Fed Chair rather than pushing to cut rates over the predicted six-month-to-one-year window.
Within four to eight weeks (from late January 2026), DeepSeek, Anthropic, and OpenAI will release new frontier AI models trained on Nvidia Blackwell servers that are significantly more capable than prior generations.
“This year, like we're going to see the first models over the course of the next four to eight weeks out of DeepSeek, out of anthropic, out of OpenAI that are trained on Blackwell servers, right? You're going to see a next generation of models far more capable.” View on YouTube
Over the next few years (by approximately 2029-2030), AI compute electricity efficiency (per token of output) will improve by roughly 70x to 100x due to changes in chip and model architecture.
“I do think that we will see compute efficiency scale by probably on the order of 70 to 100x over the next few years, meaning electricity efficiency per token of output.” View on YouTube
The predicted multi-year (roughly 2029-2030) compute-efficiency horizon has not yet elapsed.