Underclassic   rss   archive

Issue 28: Hyperscalers Spent 102 Percent of Cloud

Permanent copy of issue 28. Today's issue is always the front page.

Yahoo Finance reported that major cloud providers spent 102 percent of their cloud revenue on AI-related capital expenditure last quarter. The outlay is financed by debt and equity raises tied to new training clusters.

Date: 2026-08-29
Updated: 2026-08-29 02:00
Status: LIVE


Broadcast

Yahoo Finance reported that major cloud providers spent 102 percent of their cloud revenue on AI-related capital expenditure last quarter. Missouri utilities warned that transformer shortages will delay new data-center builds. Apple released the M5 Ultra Mac Studio with up to 512 GB unified memory for local inference. Meta published Muse Glimmer, a 30B open-weight model that runs on single consumer GPUs. Texas regulators ordered a pause on new data-center approvals pending grid studies.

Watch


Stories

Hyperscalers spent 102 percent of cloud revenue on AI capex
Yahoo Finance reported that major cloud providers spent 102 percent of their cloud revenue on AI-related capital expenditure last quarter. The outlay is financed by debt and equity raises tied to new training clusters.
The spending shows frontier training remains anchored in central clusters.
2026-08-22 / Yahoo Finance / chip / central

Missouri grid faces costly upgrades as data centers drive demand
Missouri Independent reported that new AI data centers are forcing grid upgrades and higher electricity costs for residents. Utilities warn that transformer shortages will slow further builds.
Energy constraints reinforce the central-cluster model for large training runs.
2026-08-27 / Missouri Independent / energy / central

Apple ships M5 Ultra Mac Studio for on-device AI workloads
Apple introduced the M5 Ultra Mac Studio with up to 512 GB unified memory and 80-core GPU. The system targets local LLM inference and fine-tuning.
The release widens the path for local AI on high-end consumer hardware.
2026-08-25 / Apple / chip, efficiency / local

Meta releases Muse Glimmer 30B open-weight model for consumer GPUs
Meta published Muse Glimmer, a 30B open-weight model designed to run on single consumer GPUs. Early tests show inference at 20-30 tokens per second on RTX 4090 cards.
Open weights lower the hardware bar for local model deployment.
2026-08-26 / Campus Technology / efficiency / local


Axis
local [#################-------------] central
Score: 58   clusters still winning   +1 toward central labs
Grid and capex headlines show hyperscalers continue to fund frontier training. Apple silicon and open weights provide local options but do not yet redirect the main spend.

Forces

ForceMeterScorePullNote
CHIP / CAPEX
..
76 central Hyperscalers spent 102 percent of cloud revenue on capex in the latest quarter.
ENERGY
..
84 central Grid upgrades and transformer shortages now pace new data-center buildouts.
EFFICIENCY
..
37 local List token prices keep falling and open-weight releases widen on-device options.

Line charts (0=local 100=central 2026-08-14 to 2026-08-29)

Axis score line chart

Forces line chart

Market overlay (2026-07-31 to 2026-08-29)

Latest: BTC $77,633 / hashrate ~893.3 EH/s / token markers avg $5.3033/M / axis 58. These lines are different units. We compare timing, not dollars-to-hashrate.

Market overlay line chart

Not shown: GPU cloud $/hour, hub electricity, miner hashprice, or lab cost per token -- none of those are a clean public feed.

Power and list prices
BTC $77,633 / hashrate ~893 EH/s. List-price gap (not a subsidy %): 44.4x list gap (blended) GPT-5.6 Luna -> Claude Fable 5 ($0.45/M to $20.00/M). (mempool.space + Coinbase/CoinGecko)

Marker prices (USD per 1M tokens; verified 2026-08-19; blended = (3*input + 1*output) / 4; -- = open/self-host or no single public list)

#ModelOrgCCWeightsTierInOutBlend
1 Claude Sonnet 5 Anthropic US closed
undisclosed
workhorse $2.00 $10.00 $4.00
2 GPT-5.5 OpenAI US closed
undisclosed
flagship $5.00 $30.00 $11.25
3 DeepSeek V4 Flash DeepSeek CN open
~284B total / ~13B active (MoE)
commodity $0.44 $1.32 $0.66

Full 20-model board


If useful, support the broadcast.
bc1qfs3yw5qlq8sxs50crzh3g84ug27gvww9npu85n
BTC QR

home  |  rss  |  archive  |  models  |  data  |  board  |  mcp  |  llms.txt