Le Shrub has nailed it once again... I am deep into this arena professionally and what everyone has seen is only a snowflake on the tip of an iceberg. Almost all traditional organizations are absolutely gobsmacked with confusion on it (what is it? should we? how can we benefit?, etc.) and are now paying eye-watering $s for just discussion services (in the past 3 weeks I have 4 new contracts where I got to choose who, how much and how often I let them involve me). Ferraris rolling off the assembly line, only a few driving instructors, and almost everyone still wondering if they should get a Learner's License while the race has already started or simply playing Fast and Furious in the arcade for fun
It’s funny because I heard Gavin speak at MS a few weeks ago and I came away convinced we are back in a bubble. AI isn’t new and yes it’s real but I’ve seen nothing that makes me think it’s anything other than a cap ex bubble over the next 3 years. As having some knowledge of the privates space and motivations at the moment I can tell you the underwater part is the problem. There’s no way they can get out anywhere close to 2018-2021 late stage marks and that’s the problem they are still mostly barely profitable or burning cash and that won’t change dramatically enough to offset the huge valuations that they raised their last capital at. I see a lot of private zombies for the foreseeable future. Similar to 2000s
Between VC and PE funds there is enough liquidity to keep things going on the private side - especially as they all roll their funds every 3-4 years. Add a re-commitment rate from investors of 70-90% (as assets which require duration or is duration-indifferent from pension funds / swf is ever-growing) and a huge chunk of private credit willing to lend (which also roll their funds with similar re-up rate) and there is enough private capital to keep it going for a long time. As long as the majority of underlying firms in VC and PE portfolios can continue to raise capital, find a willing private buyer or deliver free cash flows, then the show will continue to roll. With less IPO exits, the overall returns will probably be lower, in aggregate, than back in the day, but volatility will also be much much lower, so underlying investors are probably fine with that.
Thx Shrub, they actually have a very valid point about CPU and datacenter infa , if you ever work with google colab you can pick the fastest NVIDIA GPU our there to do the training but system often blows up because of CPU or RAM explodes and GPU sitting almost idle at 10% of utilization. Companies who can optimize that could be winners. And mb soon enough we will need less NVIDIA and more good old INTEL and AMD CPUs to scale it.
Rather than focusing on top down signs to assess where we are in the cycle, better go bottom up. It's like in 2022 people were looking for a "capitulation" to decide that it's a bottom..And the capitulation never came.
Great read as always. I don't think the demand for companies like CIEN will be driven primarily by AI. The argument that the infrastructure will need to keep up with the GPUs is wrong imo. The infrastructure will need to keep up with the demand for infrastructure, which makes the progress more quantitative rather than qualitative in contrast with the demand for GPUs. Kinda like - you can easily run LLM on a high end GPU installed on a 10yo system with no bottlenecks other than the GPU itself.
IPO's are in fact getting in channel for offering. ARK usually has bad market timing w/ its holdings, so while it has lagged QQQ so far in last few months FOMO momentum + its underlying holdings should push it up more before the ensuing long duration rate increases start to affect rational equity buyers (discounting "terminal value"). Then upon technical indications of top forming ARK could be a good ride down, although likely not as good as last big decline.
Also time to make list of issues having leverage & poor growth for the anticipated ride, when turn comes. Separately, even though my Dec '23 sell call on Heng Seng has seen a little movement, it likely has more potential as there are more after effects from currency & rate ripples to play out over next several months.
Interesting perspective
Le Shrub has nailed it once again... I am deep into this arena professionally and what everyone has seen is only a snowflake on the tip of an iceberg. Almost all traditional organizations are absolutely gobsmacked with confusion on it (what is it? should we? how can we benefit?, etc.) and are now paying eye-watering $s for just discussion services (in the past 3 weeks I have 4 new contracts where I got to choose who, how much and how often I let them involve me). Ferraris rolling off the assembly line, only a few driving instructors, and almost everyone still wondering if they should get a Learner's License while the race has already started or simply playing Fast and Furious in the arcade for fun
It’s funny because I heard Gavin speak at MS a few weeks ago and I came away convinced we are back in a bubble. AI isn’t new and yes it’s real but I’ve seen nothing that makes me think it’s anything other than a cap ex bubble over the next 3 years. As having some knowledge of the privates space and motivations at the moment I can tell you the underwater part is the problem. There’s no way they can get out anywhere close to 2018-2021 late stage marks and that’s the problem they are still mostly barely profitable or burning cash and that won’t change dramatically enough to offset the huge valuations that they raised their last capital at. I see a lot of private zombies for the foreseeable future. Similar to 2000s
The "max stupid" isn't a 40 P/E. It's a $3 trillion market cap, fully mature company with a 40 P/E that's "max stupid".
Between VC and PE funds there is enough liquidity to keep things going on the private side - especially as they all roll their funds every 3-4 years. Add a re-commitment rate from investors of 70-90% (as assets which require duration or is duration-indifferent from pension funds / swf is ever-growing) and a huge chunk of private credit willing to lend (which also roll their funds with similar re-up rate) and there is enough private capital to keep it going for a long time. As long as the majority of underlying firms in VC and PE portfolios can continue to raise capital, find a willing private buyer or deliver free cash flows, then the show will continue to roll. With less IPO exits, the overall returns will probably be lower, in aggregate, than back in the day, but volatility will also be much much lower, so underlying investors are probably fine with that.
Thx Shrub, they actually have a very valid point about CPU and datacenter infa , if you ever work with google colab you can pick the fastest NVIDIA GPU our there to do the training but system often blows up because of CPU or RAM explodes and GPU sitting almost idle at 10% of utilization. Companies who can optimize that could be winners. And mb soon enough we will need less NVIDIA and more good old INTEL and AMD CPUs to scale it.
“Marked safe from Mark to Market” 😂 Love it Shrub!!
Don’t bet against the Shrub and Druck
Rather than focusing on top down signs to assess where we are in the cycle, better go bottom up. It's like in 2022 people were looking for a "capitulation" to decide that it's a bottom..And the capitulation never came.
As for the Mag 7 names with the revenue growth you quoted, it's an Ad Spend rebound and A.I.fication: https://www.philoinvestor.com/p/downside-at-nvidia-and-the-nasdaq
good morning LE SHRUB
the cherry You speak of has been missing since inception ..
the remake, reboot, “HD AI” will be introduced during the next most inauspicious day they can find ..
they keep trying to infuse AI with “ethics” and they’re not quite accepting of any of the reboots that have been seen to date 🪄💎🧩
but they haven’t put a stop to them
implementing what they have on us
unsuspecting humans
Great read as always. I don't think the demand for companies like CIEN will be driven primarily by AI. The argument that the infrastructure will need to keep up with the GPUs is wrong imo. The infrastructure will need to keep up with the demand for infrastructure, which makes the progress more quantitative rather than qualitative in contrast with the demand for GPUs. Kinda like - you can easily run LLM on a high end GPU installed on a 10yo system with no bottlenecks other than the GPU itself.
Great piece Mr. Shrub (once again) - must read.
but can it be 'max stupid' with rates not zero. i am in disbelief
I am not complaining, your generosity is outrageous.
Nicely summarized 😉
IPO's are in fact getting in channel for offering. ARK usually has bad market timing w/ its holdings, so while it has lagged QQQ so far in last few months FOMO momentum + its underlying holdings should push it up more before the ensuing long duration rate increases start to affect rational equity buyers (discounting "terminal value"). Then upon technical indications of top forming ARK could be a good ride down, although likely not as good as last big decline.
Also time to make list of issues having leverage & poor growth for the anticipated ride, when turn comes. Separately, even though my Dec '23 sell call on Heng Seng has seen a little movement, it likely has more potential as there are more after effects from currency & rate ripples to play out over next several months.