E114: Markets update: whipsaw macro picture, big tech, startup mass extinction event, VC reckoning
2023-02-04 spoken.md · speaker-labeled ▶ watch ← E113 all episodes E115 →
Every number here is replayed from score_events — the same ledger the pool ranks on. Decay is what the 110 ideas nobody mentioned gave up this week; it applies only when an episode is processed.
Tier crossings
conviction thresholds crossed by this episode — 65 / 45 / 15 · ideas born here show where they landed. why 65 isn't always green
Who moved the board
each voice's force on conviction this episode — supports and opposes, weighted exactly as the replay applied them · share = % of this episode's movement
What got argued (11 ideas)
ordered by how hard each idea moved · quotes are verbatim from the transcript, timestamps deep-link into the episode
Latent startup mortality built up through ZIRP - companies that should have died from failing to raise instead kept raising - and it all comes due when the 2021 cash runs out. 50-60% of the 2019-2022 venture cohort dies in the second half of 2023 and into 2024, with down rounds, recaps and wipeouts the norm rather than the exception, because most of these companies are already worth less than their own preference stack.
Now the problem is that so many of these series B, C, D and E companies have a true market value. They're not a valueless company, but the true market value of them is probably less than the total preference stack of the capital that's gone in. So they're worthless.
I actually think this tweet storm is basically correct is, you know, I've referred to on this show before that I think one of the things that built up during this bubble is latent startup mortality. So many startups that should have died from not being able to raise the next round lived because they're able to raise money.
Sacks' call: every venture-debt underwriting model was calibrated during the ZIRP window, when startup mortality was artificially suppressed by easy equity. As the deferred mortality arrives, covenants trip, lenders pull runway early to get paid ahead of the equity, and realised loss rates on venture lending come in far worse than the models predict - a bad deal for founders and a mispriced book for the lenders.
So look, I just wonder what I don't trust is whether the return models on venture debt that were created over the last five to 10 years will be a good predictor of what the returns will be in the next five to 10 years when a lot of the mortality that should have happened in the past now happens in the future.
the venture business of the last 15 years, everyone since 2008, everyone's been trained, and all the younger people that have come up and are now partners and running the firms on an environment of momentum investing, rather than fundamental investing. And so there is also a question of how fit the investors are for a market space where valuations are flat or descending
Because we went through a phase where if you were a VP of product or a VP of design or a VP of engineering at a well-known startup, that was the most obvious onboarding into a venture firm. But if you just look back at the data, that cohort of people has actually never made money.
I'm seeing Series A and Series B companies that literally don't understand their own accounting. And so when we start talking to their accountants, there is a huge gap between what the accountants think of this business and what the founders think of these businesses.
And I think probably where we're sitting at this moment is you'd have to say that the risks of inflation returning are slightly higher, but the risks of a recession are slightly lower because with this kind of jobs report, better chance of having a soft landing here.
And that the implication of a tight job market is wages go up and wages go up, inflation goes up and because companies need to charge more because they have to pay more to get talent. And this obviously continues to support the escalatory spiral that drives inflation.
there is this friggin tidal wave of AI companies and there is this incredible amount of lubricant in the dry powder that's sitting on the side of the market right now that all these venture funds raised in the last two years that is going to lubricate all these AI companies into every vertical in every market.
Now, at the same time, I do think that if AI is the next wave, and the next sort of platform opportunity, as we all think it is, I do think that places more of an emphasis on technical skills and I was, I was literally just having this conversation at Kraft that gee, maybe the next hire we should make at Kraft should be someone who's really deep technically so they can help go deeper on technical due diligence of AI companies.
And long story short, he didn't. That was a head fake, and the markets just ripped higher. Then we went into the COVID pandemic and all of that stuff happened. So I think we're about to replay a little bit of that, at least in the next 30 to 90 days.
The tech ecosystem is clearly going through a reset and a recession. Job cuts are now the rule, valuations are much lower, whereas in the overall economy, we saw jobs support today of over 500,000 new jobs. So, the fact of the matter is that even if the overall economy avoids a recession, that doesn't mean that things are just going to bounce back. Tech is a boom bust cycle and we had a phenomenal 10 years of boom.
So these are like really interesting trends that I think just say that as rates creep back up, and if we can intersect that with some improvements in technology over the next five to 10 years that we've all talked about, it could be a real boon for startups and startup investing.
And they basically started doing some rifts and basically just getting much more efficient in what they're doing and specifically taking out layers and layers of middle management. I mean, that was really the big thing. So they kind of took a page out of Elon's book in terms of what Elon had done at Twitter.
And so, Facebook now, you can credibly see a path where Facebook could chunk out hundreds of billions of dollars of total shareholder value returned over the next four or five years. And so, for value investors, it's somewhat of a kind of a no-brainer. I mean, nothing's a no-brainer, but really, really attractive value fundamentals right now.
Episode digest
written during extraction and stored in data/extractions/ep114.json — the auditable source of truth, including everything market-adjacent that did not earn a capture
A pure markets episode two days after the Feb-1 hike and the morning of the blowout 517k jobs print, and it lands almost entirely on existing ideas. Sacks' "whipsaw economy" frame nets to inflation risk slightly higher and recession risk slightly lower, but he still calls a good chance of recession later in 2023 and insists tech is in its own bust regardless of the macro; Jason takes the other side outright ("very hard to have a recession if people are employed"). Chamath, whose own E098 bottoming call is now printing, doubles down that the pain trade is up over the next 30-90 days while simultaneously warning that looser conditions reignite inflation and force the Fed back into hiking. Two things were genuinely new: the "mass extinction" thesis, where all four hosts converge on 50-60% mortality for the 2019-22 venture cohort concentrated in H2-2023 and 2024, and Sacks' argument that venture debt's entire loss-model history was calibrated in ZIRP and is about to be falsified as deferred mortality arrives - coined five weeks before SVB failed. Meta's efficiency pivot got the biggest conviction build of the episode: Sacks on the middle-management cull, Chamath calling it an ex-growth cash machine at a PE of 11, and Jason disclosing he bought it at $94 the day of the layoffs (now $193). Friedberg reinforced his own two-month-old generative-AI-bubble call hard ("a friggin tidal wave of AI companies") and used it as the counter-argument to Chamath's austerity-vintage study. Diarization CLEAN - all four hosts present, correct counts, nine of nine addressed-by-name tests pass, fingerprints verified; last timestamp 1:11:22 with no time gap over 3.2 minutes, so the low 161-turn count is long monologues, not truncation.