@SplinterGoggles
Most models will run for free on local hardware in 10 years.
Analysis of a tweet outlining AGI pricing tiers ($20/mo to ~$10B) with mixed reaction: Support 31.11%, Confront 32.59%. Implications, cost debate & reaction.
The pricing tiers for AGI are something like (1) $20/month, (2) $200/day = ~$75,000/year, (3) $1,000/day = ~$350,000/year, and (4) ~$10 billion. For now.
Real-time analysis of public opinion and engagement
What the community is saying — both sides
tiers deliberately extract different rakes from each seat; the strategic play is to “survive tier 3 without needing tier 4,” and sellers will likely emulate ISPs by throttling and carving higher-priced tiers.
$20/month is a personal productivity tool; ~$200/day becomes “persistent labor” that runs while you sleep; $75k/yr+ is effectively replacing salaried roles; the UX and orchestration needs change radically across these bands.
once you move from a per-user fee to infrastructure-level spend, buying decisions, ops, and ROI thinking look nothing like consumer SaaS.
a $10B tier looks like an aristocracy — exclusive intelligence buys strategic power and amplifies advantages unequally.
expect multi-to-one digital-employee/human ratios (3:1–4:1), with scaling limited by compute costs and the engineering work to manage oversight and handoffs.
if users at lower tiers can use their AGI access to climb wealth and access, the tiered system may be unstable or collapse toward fewer tiers.
healthcare shows how domain-specific infra and expertise (radiology, population health, genomics) will occupy higher-priced tiers that general models can’t easily displace.
few are doing the math to compare AGI costs with offshored employees, and token/compute economics will determine when replacement makes sense.
some expect prices to keep falling with model/hardware improvements; others expect providers to raise or ration access and for deep-pocket buyers (including intelligence agencies) to dominate top tiers.
many replies argue current systems are far from general intelligence and that claims about AGI timelines are overblown.
local models, efficiency gains and self-replicating agents will commoditize intelligence and drive margins down.
the scarcity argument shifts from models to who controls access, deployment and use.
pricing is speculative: you can’t “API‑call” more intelligence into existence if hardware and capex hit physical limits.
many expect people and companies to run capable models locally or use Chinese/open variants rather than pay high subscriptions.
real‑world examples (e.g., failing to handle an ERP import) are cited as evidence AGI-like reliability isn’t here yet.
parallels to social media: free competitors and undercutting keep most users from paying, collapsing premium business models.
some insist $200/month is a realistic floor now; others propose one‑time local fees, or say pricing is meaningless until capabilities change.
a minority suggests an AGI might already exist privately, built to prevent concentration of control.
Most popular replies, ranked by engagement
Most models will run for free on local hardware in 10 years.
AGI won’t replace people equally. It’ll amplify people unequally.
Marc, can you give us an example of what someone might be doing that uses $1k/day?
Prices will will fall rapidly… models will commoditize… as they begin to recursively optimize themselves… then they will replicate themselves driving costs down even further… margins will approach zero
I will bet money, the sub sellers will reduce the usage pipe slowly for $200 down, then break off to a $500 tier and give the AS-IS usage we have right now back but higher premium. its exactly how the ISP industry played that game. the recipe is proven to work 😂
Sir ??? Where is the AGI you keep referencing ?
Found something wrong with this article? Let us know and we'll look into it.