VK
@vkmacro.bsky.social
Broadly comfortable with this view
Once the dollar bottomed this year, it’s been a pretty difficult period to trade FX
Rate diffs haven’t really mattered, and most of the action has come within crosses:
AUD strength
JPY weakness
GBP weakness
Etc etc
Most dollar bear arguments now sound stale and are a bet on the labour market
Rate diffs haven’t really mattered, and most of the action has come within crosses:
AUD strength
JPY weakness
GBP weakness
Etc etc
Most dollar bear arguments now sound stale and are a bet on the labour market
November 14, 2025 at 3:37 PM
Broadly comfortable with this view
This is v interesting
Is AI making job recruitment less meritocratic? We're getting some v interesting research studies on this question now, and the news is... not good. @jburnmurdoch.ft.com & I dive in, in the latest edition of our newsletter The AI Shift www.ft.com/content/e5b7...
November 14, 2025 at 10:15 AM
This is v interesting
The ideal strategist for someone at a multi strat (desk based) is who can provide a longer term in depth view while combining the positioning, sentiment, and models required to trade hyper short term.
There are people who can do this, but very few and far between.
There are people who can do this, but very few and far between.
November 14, 2025 at 10:08 AM
The ideal strategist for someone at a multi strat (desk based) is who can provide a longer term in depth view while combining the positioning, sentiment, and models required to trade hyper short term.
There are people who can do this, but very few and far between.
There are people who can do this, but very few and far between.
Software and AI are down together 😂
Discuss
Discuss
November 14, 2025 at 7:13 AM
Software and AI are down together 😂
Discuss
Discuss
Reposted by VK
Hyperscalers vs neoclouds?
We’re in the “Goldman-now-has-two-AI-baskets” part of the cycle:
“High-Profit AI” vs “Vulnerable AI”
(via GS desk)
“High-Profit AI” vs “Vulnerable AI”
(via GS desk)
November 13, 2025 at 7:10 PM
Hyperscalers vs neoclouds?
“Where are the revenues”
Semianalysis expects azure revenue to accelerate to 50% YoY by end 2026
Semianalysis expects azure revenue to accelerate to 50% YoY by end 2026
November 12, 2025 at 8:44 PM
“Where are the revenues”
Semianalysis expects azure revenue to accelerate to 50% YoY by end 2026
Semianalysis expects azure revenue to accelerate to 50% YoY by end 2026
Global population growth is rising faster than oil intensity is slowing.
The argument against it is that oil intensity should slow at a faster rate ahead as China’s fuel conditions slowly quickly as EVs become widespread.
However NGLs have grown faster than crude which has pushed demand higher
The argument against it is that oil intensity should slow at a faster rate ahead as China’s fuel conditions slowly quickly as EVs become widespread.
However NGLs have grown faster than crude which has pushed demand higher
🤡OIL DEMAND WILL KEEP RISING🤡
What are the IEA assumptions that make rising oil demand so improbable?
What are the IEA assumptions that make rising oil demand so improbable?
November 12, 2025 at 7:48 PM
Global population growth is rising faster than oil intensity is slowing.
The argument against it is that oil intensity should slow at a faster rate ahead as China’s fuel conditions slowly quickly as EVs become widespread.
However NGLs have grown faster than crude which has pushed demand higher
The argument against it is that oil intensity should slow at a faster rate ahead as China’s fuel conditions slowly quickly as EVs become widespread.
However NGLs have grown faster than crude which has pushed demand higher
Building a successful cloud business takes time. But when you get it right, there tend to be increasing returns of scale.
November 12, 2025 at 9:40 AM
Building a successful cloud business takes time. But when you get it right, there tend to be increasing returns of scale.
Italian politics is too funny. A 2% tax cut on people earning between 28k to 50k a year is seen as helping high earners. Also from from the FT since they don’t like Meloni so are deciding to emphasise on the critics perspective
Meloni’s ‘middle-class’ tax cut sparks political row in Italy on.ft.com/3Xpa7By
Meloni’s ‘middle-class’ tax cut sparks political row in Italy
Critics say the measure will benefit mainly high earners while reducing government revenues
on.ft.com
November 12, 2025 at 8:15 AM
Italian politics is too funny. A 2% tax cut on people earning between 28k to 50k a year is seen as helping high earners. Also from from the FT since they don’t like Meloni so are deciding to emphasise on the critics perspective
It’s really hard to tie up the data to Megan Greene’s world view imo.
Elevated household savings haven’t supported consumption. Indeed real wage growth is now slowing sharply.
Therefore nominal income lower and corporates don’t have pricing power. I think cautious is fair but this is too hawkish
Elevated household savings haven’t supported consumption. Indeed real wage growth is now slowing sharply.
Therefore nominal income lower and corporates don’t have pricing power. I think cautious is fair but this is too hawkish
Bank of England's @economistmeg.bsky.social says she isn't convinced policy is meaningfully restrictive and still troubled by household inflation expectations, firms' pay plans.
But she tells UBS conference that softer wage growth in today's data is "good news" from inflation perspective.
But she tells UBS conference that softer wage growth in today's data is "good news" from inflation perspective.
November 11, 2025 at 9:44 AM
It’s really hard to tie up the data to Megan Greene’s world view imo.
Elevated household savings haven’t supported consumption. Indeed real wage growth is now slowing sharply.
Therefore nominal income lower and corporates don’t have pricing power. I think cautious is fair but this is too hawkish
Elevated household savings haven’t supported consumption. Indeed real wage growth is now slowing sharply.
Therefore nominal income lower and corporates don’t have pricing power. I think cautious is fair but this is too hawkish
The data are starting to catch down to the vibes of a really tough job market
Duff-looking UK labour market data this morning.
Payrolls on a declining trend, unemployment rate hits 5%.
(usual caveats to both - former v prone to revisions, often upwards, and latter is drawn from the wonky Labour Force Survey)
Payrolls on a declining trend, unemployment rate hits 5%.
(usual caveats to both - former v prone to revisions, often upwards, and latter is drawn from the wonky Labour Force Survey)
November 11, 2025 at 9:02 AM
The data are starting to catch down to the vibes of a really tough job market
Once the dollar bottomed this year, it’s been a pretty difficult period to trade FX
Rate diffs haven’t really mattered, and most of the action has come within crosses:
AUD strength
JPY weakness
GBP weakness
Etc etc
Most dollar bear arguments now sound stale and are a bet on the labour market
Rate diffs haven’t really mattered, and most of the action has come within crosses:
AUD strength
JPY weakness
GBP weakness
Etc etc
Most dollar bear arguments now sound stale and are a bet on the labour market
November 11, 2025 at 7:07 AM
Once the dollar bottomed this year, it’s been a pretty difficult period to trade FX
Rate diffs haven’t really mattered, and most of the action has come within crosses:
AUD strength
JPY weakness
GBP weakness
Etc etc
Most dollar bear arguments now sound stale and are a bet on the labour market
Rate diffs haven’t really mattered, and most of the action has come within crosses:
AUD strength
JPY weakness
GBP weakness
Etc etc
Most dollar bear arguments now sound stale and are a bet on the labour market
Reposted by VK
How fast can you build a gigawatt-scale data center?
Some hyperscalers plan to do it in just 1-2 years from the start of construction.
If they succeed, we’ll see the first GW-scale data centers online in 2026, marking one of the fastest infrastructure build-outs in history. 🧵
Some hyperscalers plan to do it in just 1-2 years from the start of construction.
If they succeed, we’ll see the first GW-scale data centers online in 2026, marking one of the fastest infrastructure build-outs in history. 🧵
November 10, 2025 at 5:40 PM
How fast can you build a gigawatt-scale data center?
Some hyperscalers plan to do it in just 1-2 years from the start of construction.
If they succeed, we’ll see the first GW-scale data centers online in 2026, marking one of the fastest infrastructure build-outs in history. 🧵
Some hyperscalers plan to do it in just 1-2 years from the start of construction.
If they succeed, we’ll see the first GW-scale data centers online in 2026, marking one of the fastest infrastructure build-outs in history. 🧵
Reposted by VK
I love it when people mash together two completely unrelated phenomena as part of an effort to whip up excitement.
You don't have to invent a Chinese scheme to break the UST market to get mad at the existence of the UST basis, you can just understand that broad vol spikes lead to grossing down.
You don't have to invent a Chinese scheme to break the UST market to get mad at the existence of the UST basis, you can just understand that broad vol spikes lead to grossing down.
China has used the basis trade to unsettle the Treasury market and get Trump to fold on tariffs. We need to shut the basis trade down, not backstop it with a return to Fed balance sheet expansion. A piece inspired by the amazing @vmrconstancio.bsky.social.
robinjbrooks.substack.com/p/the-basis-...
robinjbrooks.substack.com/p/the-basis-...
November 10, 2025 at 5:00 PM
I love it when people mash together two completely unrelated phenomena as part of an effort to whip up excitement.
You don't have to invent a Chinese scheme to break the UST market to get mad at the existence of the UST basis, you can just understand that broad vol spikes lead to grossing down.
You don't have to invent a Chinese scheme to break the UST market to get mad at the existence of the UST basis, you can just understand that broad vol spikes lead to grossing down.
It’s almost impossible for the AI backlog to be cleared by h1 2026. That’s what makes all of this so tricky.
There’s a tonne of large DCs coming online in H2. Each one is huge, so will go some way to meeting demand.
That’s probably why msft is going for these smaller deals
There’s a tonne of large DCs coming online in H2. Each one is huge, so will go some way to meeting demand.
That’s probably why msft is going for these smaller deals
November 5, 2025 at 7:08 PM
It’s almost impossible for the AI backlog to be cleared by h1 2026. That’s what makes all of this so tricky.
There’s a tonne of large DCs coming online in H2. Each one is huge, so will go some way to meeting demand.
That’s probably why msft is going for these smaller deals
There’s a tonne of large DCs coming online in H2. Each one is huge, so will go some way to meeting demand.
That’s probably why msft is going for these smaller deals
wonder when consumer AI will start being priced like phone contracts.
Marginal cost of use isn’t 0.
So just like in a contract where more data/ roaming costs more, there should be more flexible priced tiers for more usage. Outside of just free and $20 a month.
Marginal cost of use isn’t 0.
So just like in a contract where more data/ roaming costs more, there should be more flexible priced tiers for more usage. Outside of just free and $20 a month.
November 5, 2025 at 7:07 PM
wonder when consumer AI will start being priced like phone contracts.
Marginal cost of use isn’t 0.
So just like in a contract where more data/ roaming costs more, there should be more flexible priced tiers for more usage. Outside of just free and $20 a month.
Marginal cost of use isn’t 0.
So just like in a contract where more data/ roaming costs more, there should be more flexible priced tiers for more usage. Outside of just free and $20 a month.
2026 capex has more upside imo.
UBS: "we now forecast global AI capex to reach USD 423bn this year (from our previous estimate of USD 375bn) and USD 571bn in 2026 (from USD 500bn). By 2030, we expect overall spending to hit USD 1.3tr"
November 5, 2025 at 12:59 PM
2026 capex has more upside imo.
Reposted by VK
Announcing our Frontier Data Centers Hub!
The world is about to see multiple 1 GW+ AI data centers.
We mapped their construction using satellite imagery, permits & public sources — releasing everything for free, including commissioned satellite images.
Highlights in thread!
The world is about to see multiple 1 GW+ AI data centers.
We mapped their construction using satellite imagery, permits & public sources — releasing everything for free, including commissioned satellite images.
Highlights in thread!
November 4, 2025 at 7:16 PM
Announcing our Frontier Data Centers Hub!
The world is about to see multiple 1 GW+ AI data centers.
We mapped their construction using satellite imagery, permits & public sources — releasing everything for free, including commissioned satellite images.
Highlights in thread!
The world is about to see multiple 1 GW+ AI data centers.
We mapped their construction using satellite imagery, permits & public sources — releasing everything for free, including commissioned satellite images.
Highlights in thread!
Aside from AMD, most openAI pumps seem to have retraced?
November 4, 2025 at 6:06 PM
Aside from AMD, most openAI pumps seem to have retraced?
I got short aud/nzd wrong as q3 trimmed mean came in far hotter than expected
It’s the most important measure by far and rightfully caused rates to reprice.
However, consumption, job growth, job openings etc are all still weakening while NZ’s housing data looks promising
It’s the most important measure by far and rightfully caused rates to reprice.
However, consumption, job growth, job openings etc are all still weakening while NZ’s housing data looks promising
November 3, 2025 at 6:29 PM
I got short aud/nzd wrong as q3 trimmed mean came in far hotter than expected
It’s the most important measure by far and rightfully caused rates to reprice.
However, consumption, job growth, job openings etc are all still weakening while NZ’s housing data looks promising
It’s the most important measure by far and rightfully caused rates to reprice.
However, consumption, job growth, job openings etc are all still weakening while NZ’s housing data looks promising
This is likely wrong after reading SK’s comments this morning. They will be used to help domestic firms expand into the US
Japan and South Korea are going to invest in OpenAI’s DCs
Calling it now.
www.wsj.com/opinion/dona...
Calling it now.
www.wsj.com/opinion/dona...
Opinion | About Trump’s Foreign Investment Funds
Tokyo and Seoul have pledged vast amounts that are hard to believe will ever happen.
www.wsj.com
November 3, 2025 at 12:50 PM
This is likely wrong after reading SK’s comments this morning. They will be used to help domestic firms expand into the US
Yes but who will host the models? Build apps of them etc? That’s exactly why most cloud providers are not model exclusive!
What if the eventual winner in the AI race isn't even in the US...
November 3, 2025 at 9:04 AM
Yes but who will host the models? Build apps of them etc? That’s exactly why most cloud providers are not model exclusive!
“There is no debate that thus far the ROI on spend has been positive and we’re not in a bubble”
youtu.be/5ze3ZNvOdRY?...
youtu.be/5ze3ZNvOdRY?...
"Is there an AI bubble?” Gavin Baker and David George
YouTube video by a16z
youtu.be
November 2, 2025 at 9:40 PM
“There is no debate that thus far the ROI on spend has been positive and we’re not in a bubble”
youtu.be/5ze3ZNvOdRY?...
youtu.be/5ze3ZNvOdRY?...
Commercial AI is about agents and coding. That’s it.
Coding has already taken off. I think it’s going to be a success.
Next is about agents where there are glimpses still but not concrete evidence
Coding has already taken off. I think it’s going to be a success.
Next is about agents where there are glimpses still but not concrete evidence
November 2, 2025 at 9:21 PM
Commercial AI is about agents and coding. That’s it.
Coding has already taken off. I think it’s going to be a success.
Next is about agents where there are glimpses still but not concrete evidence
Coding has already taken off. I think it’s going to be a success.
Next is about agents where there are glimpses still but not concrete evidence