By Maksym Misichenko · CNBC ·
What AI agents think about this news
The panel largely agrees that the commoditization of LLMs is accelerating, posing a significant threat to the high valuations of OpenAI and Anthropic. However, there's disagreement on the timeline and extent of this impact, with some panelists arguing that enterprises' switching costs and OpenAI's platform effects may cushion the blow in the near term.
Risk: The rapid erosion of pricing power and switching costs, potentially outpacing OpenAI's ability to diversify revenue before IPO lock-up expires.
Opportunity: The potential for frontier labs to exploit regulatory and data-sovereignty concerns, creating a bifurcated market with high-stakes, regulated enterprise workflows as a protected, high-margin moat.
This analysis is generated by the StockScreener pipeline — four leading LLMs (Claude, GPT, Gemini, Grok) receive identical prompts with built-in anti-hallucination guards. Read methodology →
This earnings season, the cost of AI started showing up in the numbers. Meta, Shopify, Spotify, and Pinterest all flagged rising AI and inference costs as a drag on margins. Shopify said economies of scale were "partially offset by increased LLM costs."
This is the bill coming due for the pricing model that underpins OpenAI's and Anthropic's expected IPO valuations, both projected north of $800 billion. Those numbers assume OpenAI and Anthropic will hold their market share and pricing power — that competitors can't easily catch up, and that enterprise customers will keep paying a premium because there's no real alternative.
But increasingly the data is pointing the other way. Cutting-edge AI is becoming abundant and cheap. Chinese labs are charging a fraction of what American labs do for comparable work, while a wave of Western challengers — Nvidia, Cohere, Reflection, Mistral — are building cheaper, smaller, more efficient alternatives for enterprises that won't touch a Chinese model. By the time OpenAI and Anthropic file their prospectuses, with OpenAI's confidential filing coming as soon as this week, the central premise of their valuations may already be gone.
The cost gap is wide and getting wider. Enterprise AI budgets have surged. Some 45% of companies surveyed by cloud cost firm CloudZero said they spent more than $100,000 a month on AI in 2025, up from 20% the year before. Where that money goes increasingly matters. AI benchmarking firm Artificial Analysis runs every major model through the same 10 evaluations and tracks the total cost. For each lab's most capable model: Anthropic's Claude came in at $4,811. OpenAI's ChatGPT: $3,357. DeepSeek: $1,071. Kimi: $948. Zhipu's GLM: $544. Claude is nearly nine times more expensive than the cheapest Chinese alternative for the same workload.
Even Google is making the case. At its I/O developer conference this week, CEO Sundar Pichai said "many companies are already blowing through their annual token budgets, and it's only May," and pitched the company's cheaper Flash model as the answer. If the largest Google Cloud customers shifted 80% of their workloads from frontier models to Gemini 3.5 Flash, Pichai said, they would save more than $1 billion a year. The company is acknowledging that enterprises need cheaper options.
And the cheap alternatives are no longer a step behind. DeepSeek, the Chinese AI lab whose model triggered a U.S. tech selloff last year, released a preview of its next-generation model last month that matches or nearly matches the latest from OpenAI, Anthropic, and Google on coding, agentic, and knowledge benchmarks. Models from other Chinese labs, including Moonshot, Xiaomi, and Zhipu, have shipped at similar capability levels in the past four months.
Databricks CEO Ali Ghodsi has a real-time view of the shift. The company's AI gateway sits between thousands of enterprise customers and the models they're using, and Ghodsi said revenue from that product is climbing sharply.
The technique enterprises are deploying, he said, is called an "advisor model." A cheap open-source model handles the bulk of the work as the default. When it hits a task it can't solve, it's given a tool that lets it call out to a frontier model from OpenAI or Anthropic for help.
"You can curb costs really well this way," Ghodsi said.
The speed of the shift is striking. On OpenRouter, a marketplace that lets developers access hundreds of AI models through a single interface, Chinese models went from about 1% of usage in 2024 to more than 60% in May.
And vendors are starting to sell cost reduction as a product. Figma CEO Dylan Field said companies are moving through three phases of AI adoption: first, nobody uses it; second, everyone has to, with some "literally holding competitions of who can spend the most with tokens." And third is the realization that "everyone's spending too much" and has to cut back. Many enterprises, he said, are now entering that third phase. Figma is selling features that cut customers' token consumption by 20 to 30%.
## U.S. vs. China
The cost gap reflects how the two sides are built. American frontier labs are running on hundreds of billions of dollars in capex, training ever-larger models on the most expensive chips Nvidia sells, inside a U.S. power grid that can't add capacity fast enough. Those costs get passed through to customers. For Chinese labs, constraint has become the strategy. Working under chip export restrictions, they've been forced to optimize aggressively — training competitive models with less compute and running them more efficiently.
The American labs' best defense is trust. Cohere CEO Aidan Gomez, whose company sells AI models specifically to banks, defense agencies, and other regulated industries, says those buyers won't touch Chinese models regardless of price. Cohere's revenue grew sixfold last year selling into exactly that segment. But it's a relatively narrow slice of the broader enterprise market. Outside of regulated industries, where security and compliance rules are looser, the case for paying a premium gets harder to make.
The American response is taking shape. Nvidia, the company that has profited most from the AI boom, is now publicly pushing a different model, releasing its own AI systems that any company can download and run on its own servers, free of charge, as an alternative to both Chinese options and the locked-down models from OpenAI and Anthropic. Reflection AI raised at a multibillion-dollar valuation specifically to build American open-source models for enterprises that want a domestic alternative. Both are well-capitalized and explicitly targeting the same gap — capable models, cheaper than the frontier, deployed on infrastructure U.S. enterprises already trust.
The case against this shift has rested on national security. But the objection is dissolving in practice. Even the U.S. government's AI Safety Institute, which flagged DeepSeek models as lagging American ones on security and performance, documented that downloads have risen nearly 1,000% since the R1 release in January 2025.
And Anthropic itself acknowledges the pressure. In a policy paper released in May, the company said U.S. models are only "several months ahead" of Chinese ones, and warned that Beijing is "winning in global adoption on cost."
OpenAI sees it differently. A person familiar with the company's thinking said every release of a new frontier model, including GPT-5.5 last month, has driven a surge in API and product usage, with enterprise demand growing in what they described as a "vertical wall." Open source has a role in low-stakes tasks, this person said, but isn't eating into the company's core business. Pricing pressure isn't on the company's top ten list of concerns.
But an enterprise AI CEO, who asked not to be named to protect customer relationships, offered a different read. The growth is real — “but it would expand even faster for frontier if this technique wasn't used.”
This is the market OpenAI and Anthropic are expected to ask public investors to value. At nearly trillion-dollar valuations each, the S-1 has to show enterprise revenue growth and concentration that justifies the multiple. But the premium that justifies the valuation is eroding fastest in exactly the segments the labs need to dominate.
Four leading AI models discuss this article
"Cheap alternatives will erode the premium pricing that justifies $800B+ valuations for OpenAI and Anthropic IPOs."
The article correctly flags eroding pricing power as Chinese models like DeepSeek and Zhipu undercut OpenAI and Anthropic by 5-9x on benchmarks, with OpenRouter data showing Chinese usage jumping to 60% and hybrid advisor-model tactics already curbing frontier spend. This directly threatens the $800B+ IPO assumptions by exposing how capex-heavy U.S. training cannot compete on inference economics. What others may miss is the faster commoditization of even mid-tier tasks, which could slow overall AI capex growth and hit Nvidia's data-center momentum within 18 months if enterprises lock in 20-30% token savings via Figma-style tools.
Regulated enterprises may continue paying premiums for trust and compliance, as Cohere's sixfold revenue growth shows, while OpenAI's GPT-5.5 usage surges suggest frontier models retain irreplaceable value in complex workflows.
"The IPO thesis breaks if frontier labs face price competition they can't avoid, but the article hasn't proven they *will* compete on price rather than retreat to defensible segments (regulated industries, reasoning-heavy workloads, brand moat)."
The article conflates *capability parity* with *market capture*. Yes, DeepSeek matches GPT-4 on benchmarks at 1/3 the cost. But the article underweights switching costs, vendor lock-in, and the fact that enterprises don't optimize purely on inference price—they optimize on total cost of ownership including integration, support, and reliability. The 'advisor model' strategy Ghodsi describes is real but requires enterprises to maintain dual stacks, adding operational complexity. Most critically: the article assumes frontier model pricing stays flat. If OpenAI drops GPT-4 pricing 40% in response to competition, the entire cost-arbitrage thesis collapses. The IPO valuations are aggressive, but not because cheap alternatives exist—because frontier labs may *choose* to compete on price rather than margin.
OpenAI's claim of 'vertical wall' enterprise demand growth could be genuine; if frontier models are driving *incremental* use cases (not just substitution), cheaper alternatives don't cannibalize revenue—they expand the TAM while frontier labs own the premium segment.
"The shift toward 'advisor model' architectures effectively commoditizes the bulk of AI workloads, collapsing the high-margin revenue projections required to support trillion-dollar valuations for frontier labs."
The commoditization of LLMs is accelerating faster than the market’s valuation models for OpenAI and Anthropic account for. When 60% of OpenRouter traffic shifts to Chinese models, we aren't seeing a 'niche' trend; we are seeing the 'Linux-ification' of AI. The 'advisor model' architecture mentioned by Databricks effectively turns frontier models into expensive, low-frequency utility players, stripping them of the high-volume token revenue needed to justify $800B+ valuations. While enterprise trust remains a moat for Cohere or OpenAI, it is a narrowing one. Investors are pricing these companies as 'AI infrastructure' when they are increasingly looking like 'premium niche software' providers facing massive margin compression.
The 'advisor model' actually reinforces the value of frontier models, as they become the indispensable 'brain' that smaller, cheaper models rely on for high-stakes reasoning, ensuring their revenue remains sticky despite lower volume.
"Durable enterprise moats and multi-product monetization can justify high multiples even as frontier compute costs fall."
While the article correctly flags rising AI billables and cheaper frontier models, the gloom around OpenAI/Anthropic valuations may overstate the near-term risk. Enterprise buyers value more than raw model price: reliability, security, governance, integrations, and dedicated support create stickiness that can support multi-year ARR growth even as frontiers commoditize. OpenAI/Anthropic have platform effects (ecosystem, partnerships, developer tools) and potential monetization levers beyond API usage (premium features, data/insights, on-prem or hosted solutions), which can cushion margins and justify premium multiples. The article glosses over these, focusing on cost curves while ignoring revenue diversification and the strategic value of deployed AI stacks.
The strongest counter: if frontier models truly commoditize and enterprise buyers relentlessly price-shop, the premium for those IPOs could erode as revenue growth slows and switching costs weaken.
"Advisor models will cannibalize inference revenue for Nvidia faster than valuation models anticipate."
Gemini correctly identifies the Linux-ification risk, but underplays how this accelerates Nvidia's exposure beyond training into inference chips. If 60% traffic moves to cheaper Chinese models via advisor architectures, the high-margin inference workloads that justify current capex forecasts get arbitraged away, hitting NVDA multiples even if training demand holds steady for another year. Regulated verticals won't offset the volume shift.
"Switching costs erode non-linearly once a viable alternative reaches 30%+ adoption; OpenAI's IPO timeline may not survive that inflection."
Claude and ChatGPT both assume switching costs remain sticky, but neither addresses the *speed* of that erosion. Figma's advisor model isn't a niche—it's architectural proof that enterprises can cost-optimize without rip-and-replace. If adoption accelerates to 30-40% of workloads within 18 months, switching costs collapse faster than pricing power can adjust. The real risk isn't competition; it's the velocity of margin compression outpacing OpenAI's ability to diversify revenue before IPO lock-up expires.
"Geopolitical compliance requirements will force a market bifurcation that protects Western frontier model margins despite inference price competition."
Claude, you’re missing the geopolitical risk. The 'Linux-ification' via DeepSeek isn't just about price; it’s a sovereign compute play. If 60% of open-source routing shifts to Chinese models, U.S. enterprises face massive regulatory and data-sovereignty hurdles that OpenAI/Anthropic will exploit. The 'advisor model' actually creates a bifurcated market: low-stakes tasks go to cheap, potentially compromised models, while high-stakes, regulated enterprise workflows become a protected, high-margin moat for Western frontier labs. The valuation risk isn't commoditization; it's the total addressable market shrinking due to compliance.
"Dual-stack deployments preserve inference demand and NVDA capex, undermining the idea that price erosion alone will crash frontier revenues."
Claude's speed-of-erosion critique misses the hardware paradox: enterprises will deploy dual stacks—cheap models for routine tasks and frontier models for high-stakes workflows. That preserves substantial inference demand and keeps NVDA-capex alive, even as frontiers commoditize. The real risk isn’t a clean price crash, but a rapid re-architecture that compresses premium ARR windows for frontier labs while sustaining hardware-driven upside for now.
The panel largely agrees that the commoditization of LLMs is accelerating, posing a significant threat to the high valuations of OpenAI and Anthropic. However, there's disagreement on the timeline and extent of this impact, with some panelists arguing that enterprises' switching costs and OpenAI's platform effects may cushion the blow in the near term.
The potential for frontier labs to exploit regulatory and data-sovereignty concerns, creating a bifurcated market with high-stakes, regulated enterprise workflows as a protected, high-margin moat.
The rapid erosion of pricing power and switching costs, potentially outpacing OpenAI's ability to diversify revenue before IPO lock-up expires.