CoreWeave's default swaps imply a nearly 63% chance of default within five years, up from 30%
CoreWeave's 2031 bonds yield above 16% and its default swaps trade near 1,000 basis points, Channel NewsAsia reported. At that yield a $10 billion raise would cost $1.6 billion a year in interest, three-quarters of the $2.13 billion loss analysts expect this year.
Perspective Coverage
3 publishers
- Builder
- Builder 12%
- Operator
- Operator 23%
- Investor
- Investor 65%
Reality
- Evidence72
- Adoption
- Not scored
- Hype gap+20
- Incentives60
- Confidence72
CoreWeave's bonds yield 15% when its model needs about 6%, Newcomer reports
CoreWeave's bonds yielded 15% on Thursday, nine points above the roughly 6% Newcomer says its model needs. Firmus pulled its $30 billion IPO after investors balked. Newcomer calls the $35 billion Nscale filing shaky.
Reality
- Evidence45
- Adoption
- Not scored
- Hype gap+55
- Incentives65
- Confidence50
Although CoreWeave's backlog and new commitments top 2.5 times its debt, its 2031 junk bonds yield about 16%
CoreWeave, which has never turned a profit and carries about $50 billion of debt per LSEG, has seen its shares and bonds tumble. An explainer links the slide to a wider tech selloff, though IG's Axel Rudolph puts the doubt in funding.
Reality
- Evidence58
- Adoption
- Not scored
- Hype gap+5
- Incentives60
- Confidence60
buildConfirmed2 publishers NVIDIA released the full Run:ai platform as open source, with four components that teams can adopt separately. Teams running shared GPU clusters on Kubernetes can read and modify the scheduling stack, but the license terms depend on which download they take.
Reality
- Evidence55
- Adoption
- Not scored
- Hype gap+20
- Incentives60
- Confidence65
CoreWeave lost 16% on the hot CPI print while the big hyperscalers gave up about 2%
CoreWeave's shares fell 16% on Wednesday's hot US CPI report, part of an $80 billion one-day loss across four AI stocks. A June bond it issued at a 9.625% yield recently yielded 15%.
Reality
- Evidence62
- Adoption
- Not scored
- Hype gap+12
- Incentives
- Not scored
- Confidence60
Yandex lost one of its five data centers on Thursday after a reported Ukrainian drone strike on Sasovo in Russia's Ryazan region. Running workloads moved to its other sites, but the company warned that creating new cloud resources may be limited.
Perspective Coverage
10 publishers
- Builder
- Builder 27%
- Operator
- Operator 46%
- Investor
- Investor 27%
Reality
- Evidence68
- Adoption62
- Hype gap+20
- Incentives60
- Confidence70
At least half of Nvidia's receivables now sit with five customers
Five customers each held at least 10% of Nvidia's receivables in the July quarter, up from three in January, according to a filing CNBC cited. That puts at least half of what Nvidia is owed with five buyers while the number of firms renting out its chips keeps growing.
Reality
- Evidence55
- Adoption65
- Hype gap+10
- Incentives70
- Confidence55
buildOne report1 publisher GMI Cloud offers a free week of Alibaba's Qwen and Wan models to sell its inference API
GMI Cloud is giving developers seven days of free access to Alibaba's Qwen3.8-Max, Qwen3.8-Flash and Wan3.0 via an OpenAI-compatible API. For a company built on renting GPU capacity, the week tests whether developers will take its inference service as an API.
Reality
- Evidence50
- Adoption
- Not scored
- Hype gap+5
- Incentives70
- Confidence55
Sesterce plans a 600MW AI campus at the shut Kaipola paper mill in Jämsä, Finland, with building on a 200MW first phase due this year. Buyers of European AI capacity get a site with grid and water already connected, though no date yet for booking it.
Perspective Coverage
3 publishers
- Builder
- Builder 13%
- Operator
- Operator 45%
- Investor
- Investor 42%
Reality
- Evidence45
- Adoption10
- Hype gap+30
- Incentives70
- Confidence60
buildOne report1 publisher Reflection AI commits $150 million a month to SpaceX compute before shipping its first open model
Reflection AI is preparing its first open-weight model after signing compute deals worth $150 million a month with SpaceX and over $1 billion with Nebius. Open weights leave the infrastructure and upkeep with the customer, so how the model deploys is the test buyers can run themselves.
Reality
- Evidence45
- Adoption8
- Hype gap+25
- Incentives60
- Confidence50
Nebius bought Inferize, a 10-month-old Israeli startup with 17 staff, for an estimated $100 million to $150 million, according to Calcalist. The money goes on software that keeps inference GPUs from sitting idle, and Nebius is still adding data center capacity alongside it.
Perspective Coverage
3 publishers
- Builder
- Builder 38%
- Operator
- Operator 32%
- Investor
- Investor 30%
Reality
- Evidence55
- Adoption12
- Hype gap+30
- Incentives70
- Confidence60
GMI Cloud raised $663 million: a $223 million Series B with Nvidia in the round and a $440 million credit facility from ChinaTrust Commercial Bank. Teams that need GPUs in Taiwan, Thailand or Malaysia get a funded local supplier whose expansion rests mostly on borrowed money.
Perspective Coverage
3 publishers
- Builder
- Builder 20%
- Operator
- Operator 37%
- Investor
- Investor 43%
Reality
- Evidence55
- Adoption50
- Hype gap+30
- Incentives65
- Confidence60
buildOne report1 publisher Four of five Qwen 3.8 27B engines finish within about 10% at 64 streams in g factor's tests
g factor's Qwen 3.8 27B benchmark has Together AI fastest at one stream, at 189.61 tok/s, while four of five engines finish within about 10% at 64 streams. Choosing a provider from these numbers starts with knowing how many streams the deployment will run at once.
Reality
- Evidence48
- Adoption
- Not scored
- Hype gap+18
- Incentives72
- Confidence45
buildConfirmed2 publishers New Jersey fined DataOne $1.07 million, its largest data-center penalty, for running 62 unpermitted gas generators, about 123 MW, at the Vineland AI site. Until the permits clear, that power runs on a regulator's timetable at a campus built to deliver GPU capacity to Microsoft.
Reality
- Evidence72
- Adoption
- Not scored
- Hype gap+15
- Incentives55
- Confidence68
MOUs with six large asset managers aim to mobilize more than $500 billion for AI factories. Nothing is funded yet, but the underwriting model already changes who sets compute prices.
Reality
- Evidence55
- Adoption12
- Hype gap+40
- Incentives70
- Confidence50
Contract assemblers have warned of increases above 15% on Vera Rubin and Grace Blackwell systems shipping from early next year. Every AI capacity budget written before that is now short.
Reality
- Evidence58
- Adoption
- Not scored
- Hype gap+20
- Incentives60
- Confidence60
buildConfirmed2 publishers The accelerator is in full production and the headline number is a single-request generation rate at 100,000 tokens of context. That is a different purchase order than throughput.
Perspective Coverage
3 publishers
- Builder
- Builder 48%
- Operator
- Operator 25%
- Investor
- Investor 27%
Reality
- Evidence55
- Adoption20
- Hype gap+35
- Incentives80
- Confidence60
Groq 3 LPX is in full production with Nebius as the named first customer. The benchmark is one model at one context length, and the 4x claim does not quite get from hours to minutes.
Reality
- Evidence45
- Adoption25
- Hype gap+35
- Incentives70
- Confidence55
buildOne report1 publisher Topograph turns fabric discovery into Kubernetes labels and Slurm's topology.conf
NVIDIA's open source Topograph discovers cluster topology from five named clouds or from ibnetdiscover on-premises, then republishes it as node labels, Slurm config or Slinky ConfigMaps whenever a watched part of the cluster changes.
Reality
- Evidence58
- Adoption25
- Hype gap+18
- Incentives85
- Confidence60
The AI data centre company wants $35 billion from public markets on a first half where direct costs exceeded revenue, a $1 billion net loss and a contracted-demand total of $103.4 billion. Underwriters have yet to set a range.
Perspective Coverage
7 publishers
- Builder
- Builder 12%
- Operator
- Operator 22%
- Investor
- Investor 66%
Reality
- Evidence76
- Adoption22
- Hype gap+34
- Incentives72
- Confidence74