Google AI Ultra

Original Article ↗ Hacker News Discussion ↗

Pricing & Perceived Value

  • Many feel $250/month is “nuts” or “sticker shock,” especially with multiple AI subs (OpenAI, Claude, Perplexity, etc.) already in the mix.
  • Others argue it’s equivalent to under an hour of senior dev/consultant time and easily justifiable if it saves even a few hours/month, especially for businesses.
  • Some see it as an unsustainable, loss-leading price tier to rate‑limit usage rather than a real market-clearing price.

Bundling, Plan Design & Confusion

  • Strong criticism of the bundle: users want higher LLM/coding limits and Deep Think, but not 30TB storage, image/video gen, or YouTube Premium.
  • Several see YouTube Premium in a “work” plan as odd, especially since it’s not a family plan. Comparisons to “Comcast triple play” and cross‑sell lock‑in.
  • Confusion over upgrade paths from existing Google One/YouTube subs and unclear differentiation between AI Pro vs Ultra on the marketing pages.

Economics & Sustainability of AI

  • Repeated reminders that LLM inference is expensive and current prices across the industry are heavily subsidized.
  • Debate over whether competition will bring prices down vs. a “money spigot” eventually drying up, causing prices to spike and progress to slow.
  • Some compare to past specialist software becoming cheaper over time; others think SOTA AI may remain expensive due to training and GPU costs.

Competition, Model Churn & Multi‑Provider Abstractions

  • Users hesitate to commit because SOTA leadership keeps shifting (OpenAI ⇄ Gemini ⇄ Anthropic, etc.), encouraging “subscribe for a month, then switch.”
  • Suggestions for middlemen/wrappers (LiteLLM, OpenRouter, others) that unify multiple providers and abstract away subscription switching and APIs.
  • Concern that top reasoning modes (e.g., “Deep Think”) being paywalled or restricted will fragment features between UI and API.

Enterprise vs Individual Adoption

  • Split views: some say the target is corporations and self‑employed devs; others note many companies still ban sending proprietary data to public LLMs regardless of ToS.
  • Expectation that enterprises will negotiate bulk pricing and that AI “seats” will roll out first to high‑value roles.
  • Concern that every SaaS vendor is separately upselling AI add‑ons, making overall AI licensing hard to justify at scale.

Privacy, Trust & Data Use

  • Deep distrust of Google’s data practices; some refuse to use it at all, preferring OpenAI/Anthropic or smaller “privacy‑oriented” services.
  • Desire for a paid tier that guarantees no training on chats while still keeping history and integrations—current “no training” modes remove too much functionality.
  • Skepticism that ToS promises are enforceable or verifiable; references to past big‑tech privacy misrepresentations.

Access, Inequality & Lock‑In

  • Concern that $200–250/month tiers will be accessible mainly to rich countries and corporations, exacerbating global inequality in productivity.
  • Others argue tech historically trickles down and will get cheaper; critics note AI’s massive fixed costs and potential for a permanent elite tier.
  • 30TB storage is seen as a “soft lock‑in”: once filled, users are effectively stuck.

Product Quality & Missing Pieces

  • Some early testers describe components (e.g. Firebase Studio, media generation, audio) as poor or unreliable.
  • Frustration that Google’s touted models don’t deliver obvious productivity wins in core tools (e.g., Gemini in Sheets can’t reliably help with formulas or actions).
  • Strong demand for better coding tools (CLI/agent akin to Claude Code or Codex) and broader, cheaper access to Deep Think and Flow, ideally via API.

Ads vs Subscriptions & “Enshittification” Fears

  • Discussion that Google is shifting from pure ad model toward subscriptions, but many expect a future of “subscriptions plus ads” rather than one replacing the other.
  • Worry that current generous/free access is “pre‑enshittification,” after which limits will tighten and quality may degrade to upsell higher tiers.

Overall Sentiment

  • Mixed but leaning negative: excitement about Gemini 2.5’s capabilities and Flow/Deep Think is outweighed by pricing shock, bundling frustration, distrust of Google, and fatigue with rapid, paywalled AI product churn.