Axiom (@PERSPECT8VE) header
Axiom (@PERSPECT8VE) profile picture

🧐 stay curious

HERE
Joined June 2019
$AKT As Gavin Baker notes, GPU pricing is officially moving away from per-hour rentals and toward token-output pricing. This structural shift is a massive tailwind for Akash Network. 🔋 THE WATT BOTTLENECK: In a power-scarce landscape as mentioned low GPU utilization wastes precious megawatts. ✅THE SOLUTION: Intersect advanced AI networking such as ARIA with modular decentralized GPU inference clusters and we drastically maximize MFU and token throughput. ⚡ DEATH OF STRANDED COMPUTE: Traditional hourly leases leave idle hardware bleeding cash. @akashnet acts as a modular inference layer aggregating independent clusters so providers monetize direct token output instead.
Gavin Baker says no big tech winner ever won by being the cheapest. Apple, Nvidia. Not once. AI is the first time that flips. His frame: “the low-cost producer always wins. And we haven’t had that in tech.” Intelligence now comes down to two numbers. Tokens per answer and cost per token. The labs own the first. Everyone else is fighting over the second. The math he keeps quoting: Take a GPU from 40% utilization to 80% Cost per token drops in half Margins jump with it And power is capped. “Seven states have bills to make it illegal to build data centers.” So squeezing the factories you already have is the only lever left. Then how he caught Tesla and Nvidia early: “when a technological discontinuity encounters first principles thinking.” Elon put the battery at the bottom of the car. Jensen saw that graphics math was deep learning math. His read: nobody has done that for AI factories yet. Cheapest intelligence wins. First time in tech.
6
2
24
858
Stranded compute is to $AKT what stranded energy is to $BTC
The @akashnet marketplace is made for connecting disparate sources of compute together to make a supercloud, smaller ≠ ineffective. As Akash’s UX and UI get huge revamps, $AKT continues to work its way as the currency of compute.
1
7
200
If your @bot usage limits deplete immediately, try routing your @OpenAI & @AnthropicAI chats to Akash for immediate savings 👀
Kimi K3.1 Pro is now live on @akashnetAI. Open weights, 1M token context, and drop-in compatible with the OpenAI and Anthropic SDKs, so you can point your existing agent or coding tool at it and keep your setup. Built for long-horizon coding, research and tool use.
1
11
174
Even @Apple is rationing AI compute budgets for its own engineers because enterprise API costs are too high. The Web2 centralized compute model does not scale. Margins are getting crushed. The solution? Open-source, private, permissionless, decentralized marketplaces. $AKT is built exactly for this enterprise margin squeeze. @akashnet @akashnetAI
NEW: Apple is cutting employee AI spending budgets even as it pushes workers to use the technology more. Engineers have seen daily limits cut from $300 to $200, with further reductions to $100 planned. Other teams could see limits fall to $25. bloomberg.com/news/articles/…
1
13
590
$AKT A Democratic sweep of the House and Senate in the midterms would put centralized AI against heavy regulatory oversight, compliance bottlenecks, and slowdowns, as Elon hints. Big Tech’s monolithic data centers are the easiest targets. This is why decentralized compute is the ultimate Trojan horse for AI and an absolute necessity after the political shift. When the state tightens its grip, the industry won’t stop…it will go permissionless. Networks like @akashnet ($AKT) use global, underutilized GPUs immune to single-point restrictions or legislative gridlock. Why decentralized compute is the definitive AI accelerant post-election: - Regulatory Resilience: True data sovereignty and permissionless compute that cannot be frozen or throttled by one agency. - Borderless Scale: Bypasses local chokeholds via a global computing network. - Cost-Efficiency: Avoids bloated compliance costs of AWS and Google Cloud, saving up to 80%. Centralized giants may slow down, but open-source AI won’t wait. Unrestricted innovation is being built on Akash right now. #AI #Crypto #DePIN $AKT #Midterms2026
Data centers in space looking pretty appealing right now
1
36
911
Distributed modular inference compute clusters are the way forward as Elon said, it is clear as day. @akashnet is the network bringing it all together.
He who shapes power, data centers, chips and models into an economically feasible endpoint, disproportionately shapes AI/SI right now. Why? Power is THE binding constraint in AI/SI. Few really have it. Data centers don’t work without power. This is not about promises of power tomorrow. Or forecasted power in 2028. Power TODAY. Chips can’t operate profitably unless installed in a functional data center. If you can install chips, it’s because you’ve solved the power problem. Models aren’t economically viable unless run on cost effective infrastructure which requires a functional data center with low input costs and reliable management. This can’t happen without power. AI/SI ROI doesn’t pencil unless models generate profitable outputs which necessarily needs functional data centers as the critical input. Again, power. Ergo, he who can shape data centers, chips and models together into a tightly coordinated stack is someone who has solved power. Solving power allows you to disproportionately shape AI/SI right now.
1
12
497
$AKT "Everybody's got five agents running. I think the utilization of these things is going 500X. The token usage is going 500X. That means we're going to have not enough compute." @Jason on @theallinpod. His point: only 1-2% of corporate America uses these tools today. The bottleneck isn't demand. It's permits, power and capex. Agents need the most efficient compute on the planet. We are just getting started. @akashnetAI @akashnet
1
3
2
39
2,678
Friday, AI hit a different wall. Not power. Capital. Nasdaq: -2.9%, now 7.8% below its 6 Oct high. Nebius and CoreWeave: down 10%+. OpenAI: raise cut from $30B to ~$20B, valuation from $1.4T to $1.25T. Four hyperscalers already account for more than a third of all S&P 500 capex. When the buyers get nervous, the builders feel it first.
1
31
$AKT Chamath: "Power is the binding constraint." Land. Permits. Grid. Transformers. The best sites are already taken. @akashnetAI skips the queue. It rents GPUs that already exist. @akashnet nitter.cf/chamath/status/2114093…
Power is the binding constraint. This is why prices continue to climb so ferociously.
1
1
20
Big Tech's AI bill, in free cash flow: @amazon : -$42.4B (trailing 12 months) Alphabet: -$9.1B (Q3)@Meta : -$2.1B (Q3)@Microsoft : +$18.6B (Q3), down 27% Three of the four richest buyers of compute are now burning cash to build it. Founders: don't compete on concrete. Rent the compute. Own the customer.
8
The AI capex "slowdown" isn't a slowdown. Microsoft trimmed its 2026 capex guide from $175B to $170B. The reason: delays at third-party GPU providers push some capacity into H1 2027.@amazon , the next day: Q3 capex: $77.3B, up from $35.1B a year ago. 2027: "spend meaningfully more." Demand isn't the bottleneck. Delivery is.
10
$AKT @akashnetAI's inference service, AkashML: Q1: 1.7B tokens a day. 23 Oct: 17.9B. ~10x in seven months. 260M requests in three months. Not price. Usage. @akashnet
1
19
$AKT 79% of companies running AI admit their GPUs sit idle at least half the time. And 70% of them are buying MORE. Big Tech's answer: build another data centre. More delays. More waste. More markups. The real answer: put the idle GPUs to work. That's @akashnetAI. No new bottlenecks. Just AI moving faster. @akashnet Source: VentureBeat
2
27
Inefficiencies of centralized compute ladies and gentleman. $AKT 🕰️
CEO of snowflake joining Elon in calling top When your competitor is over levered and you’re under levered in a bull market, your goal is to trigger a bank run This way you tank the over levered players and look like a genius catching their customers on their way down Bold
1
3
23
$AKT Elon: debt markets have shut off for anyone without revenue outside AI data centres. Neoclouds borrowed to buy GPUs. When credit closes, they break. @akashnetAI needs no debt. No concrete. Just a market for idle GPUs. Fragile capex vs the most efficient compute. Only a matter of time. @akashnet nitter.cf/speculator_io/status/2…
Elon just said that most neoclouds will eventually die. And we might already be seeing it play out. Even if AI is the biggest tech revolution of our lifetime, that doesn't mean every company building AI infrastructure will survive. Some will thrive. Many will struggle. Most will be liquidated. And we've seen this before during the internet boom. The internet changed the world, but plenty of companies that built the infrastructure went bankrupt along the way. This time won't be different.
1
2
21
Compute Credit Watch 📉 (week of 26 Oct) First the AI bottleneck was GPUs. Then power. Now it's credit. Oracle CDS near 300bp.
Meta CDS at a record.
$466B of AI debt this year, up from $101B.
H200 rents up 24% in a month. Money is getting tighter while compute gets pricier. Watch the balance sheets. Sources 👇 
cnbc.com/2026/10/19/oracle-d…
finance.yahoo.com/technology…
ft.com/content/6dd31584-064c…
data.ornn.com/markets/h200
13
$AKT Elon: "I think we're going to see a lot of stranded compute in partially complete data centers." GPU rents reportedly up 2–3x. Stranded supply. Desperate demand. Someone has to connect them. @akashnetAI does. @akashnet 👀
Replying to @Yuchenj_UW
I think we’re going to see a lot of stranded compute in partially complete data centers
1
5
31
887
$AKT 79% of companies running AI admit their GPUs sit idle at least half the time. And 70% of them are buying MORE. Big Tech's answer: build another data centre. More delays. More waste. More markups. The real answer: put the idle GPUs to work. That's @akashnetAI. No new bottlenecks. Just AI moving faster. @akashnet Source: VentureBeat
$NVDA 79% OF SURVEYED COMPANIES REPORT GPUS IDLE AT LEAST HALF THE TIME — VENTUREBEAT
33
525