Acetaldehyde accumulater; circle of competence ai, scaled economies shared and platform aggregators
New Zealand
Joined November 2020
- Tweets1.5K
- Following717
- Followers304
- Likes15K
Jackson Matysik retweeted
$NBIS $IREN $CRWV
-GPU depreciation went from 4 years to 7-8+ years:
the market didn't care
-GPU rental prices rocketed like crazy:
the market didn't care
-Payback period went from 3 years to 1 year:
the market didn't care
-Cost of capital increases by few percent
OMG THEY ARE GOING BANKRUPT SELL ALL THESE S*HT
🟢1, You can have a high cost of capital
🟢2, As long as the payback period is short you can have an incredible ROIC
The last 12 months we've been in a market that didn't reward the "neoclouds" for substantially bigger good news, but it instantly punished them for the substantially smaller bad ones.
Jackson Matysik retweeted
$CRWV You asked why Goldman upgraded it. The answer was in the note, which you clearly didn't read.
The "pump":
• One upgrade. Goldman's Gabriela Borges moved CoreWeave to Buy and cut her target from $139 to $106, 24% lower. Nobody pumps a stock by cutting the target by a quarter
• Her reasoning: the scary bond yields and CDS levels apply to the unsecured debt, most of the financing is asset-backed loans secured by GPUs and customer contracts, and she sees CoreWeave's cost of capital rising by 100-200 bp at most
• The stock closed down 0.44% at $61.12. If that was a coordinated pump, someone should get fired
Liquidity, straight from the Q2 10-Q (June 30):
• $5.5B of cash, $6.9B including restricted cash
• ~$10B of signed, undrawn facilities including a $2B revolver, for ~$15.6B of total liquidity
• Operating cash flow of +$3.66B in the first half of 2026, against -$190M a year earlier
• Customer prepayments sitting in deferred revenue rose to $9.7B, from $8.2B in December
• $103.7B of contracted backlog, 41% of it scheduled to be recognized by June 2028. That's ~$42B of revenue already signed for the next two years
• Q2 adjusted EBITDA of $1.51B covered $640M of net interest ~2.4x
The debt you're scared of:
• $35.6B of principal at June 30. $4.4B is due in the rest of 2026, $6.2B in 2027, and $14.9B after 2030
• 38% of it is delayed draw term loans secured by GPUs and contracts, generally with investment grade customers. They amortize from those customers' payments and the lender is paid first, which is why so much of the near-term schedule pays itself
• Only $10B is unsecured senior notes, the slice behind every 15% headline. Its 2032 notes went from a 15.75% to a 14.78% yield today, and the CDS that spiked above 1,000 bp last week has already eased
• $3.7B of the debt is non-recourse
• Per the filing, every 1-point move in rates changes quarterly interest by ~$30M. The top of Goldman's range, +200 bp, is ~$60M a quarter, against $1.51B of quarterly adjusted EBITDA
The market that supposedly won't fund it:
• In Q2 alone it closed a $3.1B term loan, the first publicly syndicated delayed draw facility backed by HPC infrastructure, raised more than $10B of unsecured debt and convertibles including its first Eurobond, and took $1B of equity from Jane Street
• On September 22 it closed $4.2B of convertibles at 2.875% due 2033, upsized from a $3.0B launch. Its April convertibles carry 1.75%
• That's more than $18B raised since April. Companies "in a lot of trouble" don't upsize a convert at under 3% four weeks before your tweet
• Q2 funding was $13.5B of net debt proceeds against $1B of equity. CoreWeave doesn't need a high share price to fund itself, so the "pump it for liquidity" theory falls apart on its own
The customers and the backstop:
• Backlog is $104B, up 246% YoY, plus $25B+ of new commitments signed early in Q3
• Its largest customer fell from 72% of revenue in the first half of 2025 to 40% in the first half of 2026. The customer base is broadening, with OpenAI contracts totaling $22.4B and a Meta deal worth up to $14.2B
• Nvidia owns ~10% and signed a $6.3B order obligating it to buy CoreWeave's unsold capacity through April 13, 2032
• Q3 revenue is guided at $3.45-3.6B, up from $2.58B in Q2. 2026 is guided at $12.4-13.2B, and exit-2026 ARR at $18.5-19.5B
Valuation:
• $33.7B market cap against a $104B backlog, ~2.6x this year's revenue guide
• The average target across 41 analysts is $140.58
Liquidity is the one thing CoreWeave keeps proving it can get, most recently $4.2B at 2.875% four weeks ago. Read the filing before you tweet about it.
Why are they trying to pump $CRWV w upgrades all day and on CNBC?
Liquidity . . . this company is in a lot of trouble.
Jackson Matysik retweeted
Just like when the world realized, with the arrival of the iPhone that maybe you would, as your teachers disputed, “always have a calculator in your pocket”, I’m beginning to realize I will never have to be good at doing my digital chores again.
I can relax about my complete ADHD-driven incompetence in filling out my taxes, or responding to emails, or paying my bills, as I will always have an executive assistant in my pocket. A new era is here
Jackson Matysik retweeted
> GREETINGS, PROGRAM.
> WE'VE BEEN EXPECTING YOU.
lemonade.com/agentic
Jackson Matysik retweeted
$CRWV isn’t paying 15% interest. Only 4% of their debt costs that.
I went through the Q2 10-Q line by line. Out of ~$35B in debt, exactly one loan sits at 15%. It’s $1.3B, due March 2028, and already being paid down.
The 15% chart going around is the market yield on their 2032 bond. A higher yield doesn’t change what CoreWeave pays. It works like a mortgage. If you locked in at 3% and rates go to 7%, you’re still paying 3%. CoreWeave locked that bond at 9.625% and still pays 9.625%.
What they actually pay
> $6.6B of convertibles at ~2%
> ~$10B of bonds at 9-10%
> Most secured loans around 9%
> ~8.4% blended across everything
They have roughly $20B of 2026 capex left to fund, and 2027 will need more. Their main funding tool has been loans backed by customer contracts. The investment grade one backed by Meta priced around 5.9% in March. If they keep borrowing that way, I think the math holds.
Zooming out, the AI trade still looks intact to me. $ASML reported record Q3 profit this morning and said lithography demand will outrun their supply in 2027 and 2028, even after adding ~30% capacity each year. Customers are already asking about 2029 and 2030 slots.
Demand is still outpacing supply everywhere I look. $CRWV $ASML
Jackson Matysik retweeted
Yes.. let’s shall we!? 🤦♂️
$LMND launches since July 7th…
• Renters: VT, ME, NE, KS, KY, AK, HI - 7 states
• Car: FL, MO
• Tesla FSD: TN, MO
• Pet: KY
• HW3/FSD Lite: AZ, CO, TN
All before Investor Day. I mean c’mon…they’re more than likely holding back some announcements for the event 👀
Risks… and stuff I would love to have more clarity on…
1. Software has been commoditised - how will you use this to your advantage? - not just moving faster on existing product launches
2. How do they plan to scale and become the agent of choice provider when they simply can’t quote for 100% of consumer asks (fullstack with capital limitations) - will they shift to (MGA / Aggregator - for all policies they can’t self insure?) it’s important Insurify and others don’t continue to gain ground here.
3. Potential Long term shift from personal to commercial lines - how will they navigate this?
4. What will they do to expand margins past what is typical in Insurance? - holding company, Ai Incubator, make it pointless for others to bother using Ai to rebuild their stacks by white labelling their own… (approved by regulators, compliant, ISO 27001 and PCI compliant) - there are many reasons startups in this space would prefer to leverage their stack - the stripe for Insurance or more than this - the AgentOS for anything…
If OpenRouter was sold for $8bn - then LMND should be worth sooooo much more… why? Because routing is a tiny fraction of what they built. Routing, testing and agentic QA, CS, regulatory compliance and risk predictions with 10 years of private proprietary data!.. BUT they have to show the net income being generated at the high margins to benefit from said platform.
That graph, Infastructure and connected platform - can be used for SO much more than Insurance.
Institutions continue to accumulate while retail sell and moan…Story as old as time. 🤦♂️
They are playing this right, growth at almost all costs. Redistribute budgets into product and development, not just using Ai to deliver the same or a little more while cutting engineering heads.
People ONLY JUST now understanding agents and their power with the launch of muse and others… and freaking out while these guys have been using them for years, building the infrastructure, the architecture to be able to navigate the data in a fast, deterministic, compliant manner.
One of the main reasons I bought LMND was seeing the Graph they built all those years ago… an architecture that’s even more important today than it was then….
99% of people will not understand the importance and implementation of this in a truly agentic world we rapidly approach.
The things that take me the longest now are data related… I’m often in disbelief how far and fast things have advanced - I typically have 20-70 agents running at any one time - it can write code so incredibly fast… yet I’ve been at a bottleneck on worktree maintenance and institutional ownerships reconciliation and visual QA - weeks and weeks of agent runtimes, why? - because physical read/write latency has limits, and managing hundreds of parallel worktrees creates storage challenges / storage has limits and it was three areas I underestimated the complexity and scale requirements - and now it’s painful… but this is the EXACT pain (honestly incumbents will be so much worse because they have to do it with live customer data and vastly more of it) the large incumbents will face when trying to shift from their standard BAU process.. it’s harder to go as fast as agents now allow than people realise yet…. IMO this IS their moat, this IS their advantage…
The winners of tomo will be those that built underlying systems for unthinkable automated scale, have proprietary data and or physically produce something.
We are so early.
Jackson Matysik retweeted
Why do nihilists who talk about humans being specks of dust on a space rock in the cosmos always zoom out but never zoom in? Our bodies are universe-like structures for microorganisms, who are universes for particles. We are incredibly massive and significant
Jackson Matysik retweeted
On Diary of a CEO I said plumbers were safe from AI. I was wrong.
Today I woke up and there was no hot water. I took a photo of my boiler, the error message and the pipes. Gave it to ChatGPT.
It walked me through exactly what to do and I fixed the problem. It wasn’t too complex but it could have easily been something I would have called a plumber for previously.
In the future if I had AI glasses a wider range of DIY projects would be in grasp. Either for me or for anyone inclined to be a local handyman.
I’m sure a significant portion of plumbers currently do call outs for fairly benign tasks.
All jobs of the future will be transformed by this technology. The way we are going the most common job title in the next decade will be “founder” and “cofounder”.
Jackson Matysik retweeted
"I'm a plumber, robotics won't catch up for another decade or two, I'm safe"
Oh yeah? What about when the supply of plumbers multiplies 10x because it's one of the few buoyant occupations? You better be the Gordon Ramsay of plumbing
Jackson Matysik retweeted
Math was bound to fall first. The purer the field the easier it is for AI to crack.
Jackson Matysik retweeted
“okay, so what? it’s just math.”
no, you don't get it.
here’s what OpenAI’s math breakthroughs could eventually help make possible in the real world:
Vlasov–Maxwell → stronger foundations for fusion research, whose ultimate prize is virtually limitless clean energy to power civilization
Calderón’s Problem → portable body scanners using electrical signals, making medical imaging cheaper and easier to access
Maximum-Cardinality Matching → faster searches for compatible kidney swaps across large donor pools, helping hospitals coordinate lifesaving transplants
Inverse Elasticity Problem → scans that map tissue stiffness, helping doctors locate suspicious growths inside the body
Mumford–Shah Conjecture → better tools for spotting tissue changes in brain scans, helping doctors identify signs of disease
Bose–Einstein Condensation → stronger foundations for quantum sensors that could help vehicles navigate without GPS
Simple Stochastic Games → better safety checks for self-driving cars and robots before dangerous mistakes reach the real world
Matrix Multiplication → cheaper AI and bigger scientific simulations using the same computers
Edit Distance → faster DNA comparisons, helping researchers study genetic changes linked to disease
The k-Server Problem → robots that waste less movement and energy, making warehouses more efficient and goods cheaper to move
a reminder that math is the foundation of science.
so accelerating mathematical discovery could compress centuries of scientific progress into years.
insane timeline to be alive for!
We’re releasing a broad range of new mathematical results produced by an internal frontier model.
We’ve been consulting with the independent Advisory Group on Mathematics and Artificial Intelligence at the Institute for Advanced Study, and we have drawn on their advice and public recommendations to inform how we release these results.
github.com/openai/math
Jackson Matysik retweeted
“On average, each result used three hours of ChatGPT Pro thinking compute with that model.”
We’re releasing a broad range of new mathematical results produced by an internal frontier model.
We’ve been consulting with the independent Advisory Group on Mathematics and Artificial Intelligence at the Institute for Advanced Study, and we have drawn on their advice and public recommendations to inform how we release these results.
github.com/openai/math
Jackson Matysik retweeted
if you genuinely believe anthropic could build the machine god this valuation is totally reasonable. if not undervalued. investors are thinking about the probability distribution of future outcomes. there is absolutely 0 chance that Samsung builds the machine god. There is like a non-zero chance that Anthropic or OAI does, even if its infinitesimally small
Jackson Matysik retweeted
i don't think nearly anyone, myself included, has truly internalized there'll soon be a machine better than us in every single intellectual & physical capacity
deep down we all share a feeling that our 'entrepreneurship' or 'taste' or... is special & safe. it isn't. it's over
