THE POST
“Lack of electricity” is an oversimplification. Everything is hard about bringing massive compute online at extremely high speed. If there was twice as much time, this would be relatively easy, but there is not. https://t.co/rmuECPbCrN
QUOTED
Rihard Jarc@RihardJarc
Former $GOOGL employee on why data center capacity and electricity are the real bottlenecks, not GPU production itself, and how unreliable some of the neoclouds are ($AMZN, $GOOGL, $MSFT, $NVDA): 1. According to the expert, open-source models are closing the gap with frontier models faster than expected, with customers increasingly replacing Anthropic and OpenAI models in multi-agent architectures with open-source alternatives starting from Q2 of this year. The shift started with simple tasks like tool calling and MCP, but has moved up the stack to low-level orchestration. 2. The expert identifies capacity as the biggest constraint at the moment, with certain companies' traffic growing 5x while GPU availability, particularly in the U.S. where latency requirements are tightest, simply cannot keep up. Sourcing capacity overseas in Asia or Australia is technically possible but adds latency that most customers will not accept. 3. Hyperscalers offer reliability but are expensive, require multi-year commitments of three or more years, and have limited availability of the latest hardware, with customers already chasing B300s. AI-native companies tend to gravitate toward neoclouds due to hyperscaler pricing and contract rigidity, but neoclouds come with their own reliability issues, with GPUs dropping off unexpectedly and requiring engineering resources to manage fleet stability around the clock. 4. The third tier consists of smaller independent operators who have invested in GPU racks, seeing the opportunity to lease out capacity and recoup costs within a year or two, with the remainder being pure profit. The expert describes these as essentially unusable in practice, with operators who do nothing more than find land, install racks, connect electricity, and set up cooling, with no real cloud management, maintenance, or operational expertise behind them. 5. The expert expects overall GPU availability to increase meaningfully over the next year or two, with $NVDA continuing to ramp production. The real bottleneck is not GPUs themselves but data center capacity and electricity in the U.S. On the supply side, the expert also sees players $SPCX beginning to lease out spare capacity to the market, with more likely to follow. 6. The expert notes that U.S.-based neoclouds typically require 20-30% prepayment upfront, with the remainder paid monthly or quarterly over the contract term, and rebates negotiated at the end for any reliability incidents like traffic loss. APAC-based providers operate differently, with no prepayment required at all. found on @AlphaSenseInc
https://x.com/RihardJarc/status/2112562002958692392Text is the full long-form text the X API returned for the quoted post. It carries four images, which are not copied here. Observatory did not see the underlying AlphaSense interview.
ENGAGEMENT
- likes9,499
- reposts1,172
- replies1,437
- quotes160
- bookmarks1,087
- views1,599,397
AS RETURNED BY THE X API · FETCHED 2026-10-21 05:25 UTC
WHAT IT SAYS
Elon Musk pushes back on a framing he calls an oversimplification: that a shortage of electricity is what holds back AI compute. His view is that every part of bringing huge amounts of compute online very fast is hard, and that with twice as much time it would be relatively easy. He is quoting a post that relays an expert-network interview with a former Google employee, who said data-centre capacity and electricity in the US, not GPU supply, are the real bottlenecks.
CONTEXT
Musk posted this on the day of SpaceX's Q3 2026 earnings call. On that call, as reported by Stocktwits via Yahoo Finance, he named power, cooling and infrastructure as bigger constraints than chips, said SpaceX could rent out its compute several times over, and described finding no spare GPUs of any kind to move inference onto. SpaceX ended September with 1.7 GW of nameplate compute and expects 2.5 to 3 GW by year-end.
The electricity side of the argument has numbers behind it. Morgan Stanley estimates US data-centre developers face a 32 GW power shortfall through 2028, about 34% of projected requirements, even after counting extra generation, and says more chips are being built than there are places to put them.
Musk's broader point also has support: the bottleneck is a stack, not one item. BlockWest's October tracker counts 14.4 GW of planned IT power across eleven big US AI campuses, of which 4.1 GW was running, and points to gas turbines (116 GW of GE Vernova backlog and reservations against about 20 GW of yearly output), interconnection queues and permitting as separate gating items. Oracle and OpenAI dropped a 600 MW Abilene expansion partly because power would arrive after the GPU generation turned over.
WHAT IT MEANS FOR K
Neutral as a statement, useful as a diagnosis. Whether the binding constraint is electricity alone or the whole stack of turbines, transformers, cooling, permits and chips, the effect on K is the same: compute demand is pulling on harnessed power faster than the system can deliver it. Time, not intent, is the scarce input.
CLAIMS
- 'Lack of electricity' is an oversimplification of what limits AI compute build-outs.
- Everything about bringing massive compute online at very high speed is hard.
- With twice as much time, it would be relatively easy.
EVIDENCE
- Yahoo Finance (Stocktwits): on SpaceX's Q3 call Musk named power, cooling and infrastructure as bigger constraints than chips; SpaceX had 1.7 GW of compute at end-September.
- Economic Times (ANI): Morgan Stanley sees a 32 GW US data-centre power shortfall through 2028.
- BlockWest: 4.1 GW of 14.4 GW planned IT power was operating across eleven large US AI campuses in early October; turbine supply and interconnection are separate bottlenecks.
REASONING
- The post reframes a bottleneck rather than announcing new capacity, so by itself it neither adds nor removes harnessed power.
- The cited evidence supports both the quoted post (power is short) and Musk's wider point (many parts of the stack are short at once).
- No change to harnessed power follows from the post, so the verdict is neutral.
SEE ON THIS SITE
TAGS
- COMPUTE
- DATA CENTRES
- GRID
- POWER
- SPACEX