The Board Room
Chinese labs are pricing inference 10-28x below the US frontier and still running 50-70%
The frontier-pricing assumption underneath most AI budgets has a 6-12 month shelf life. A vendor contract signed this quarter without a cost-collapse scenario in the model is one that gets renegotiated under duress instead of by choice.
AI Cost Structure Inverts: Chinese Efficiency + Margin Collapse
Chinese labs extract 4-7x more capability per compute unit while pricing 10-28x below US competitors at 50-70% margins. Simultaneously, AI-native businesses structurally cap at ~17% gross margin vs SaaS's 70%. 4B-parameter models now match frontier performance. The premium-compute thesis is being outflanked.
Agent-First Platforms Replace the Interface Layer
Google killed ChromeOS, merged with Android, and made Gemini the primary interaction layer with 5 OEM partners shipping fall 2026. Amazon replaced its search bar with an AI agent that buys from competitor sites. Study of 16,000 AI shopping rounds: 7 of 8 traditional conversion tactics fail. The interface is no longer the product.
OpenAI Finetuning Kill Forces a Three-Year Strategy Bet
OpenAI deprecated finetuning APIs, eliminating the middle path between 'consume frontier as utility' and 'run your own weights.' The market bifurcates: top 1% (Cursor, Cognition) invest in RLFT on open models as their moat. Everyone else competes on prompt engineering — a discipline every competitor has equal access to. Straddling is the worst outcome.
Enterprise AI Adoption Bottleneck — Metrics Are Broken
Employees at Amazon, Meta, and Microsoft are gaming internal AI adoption metrics. Google, OpenAI, and Anthropic all launched enterprise deployment arms in the same week — an industry admission that implementation, not capability, is the binding constraint. PE firms (Vista $750M, Blackstone, KKR) becoming primary distribution channel.
Macro Regime Shift: Inflation Crosses Wages, Fed Paralyzed
April CPI hit 3.8% against 3.6% wage growth — first negative real-wage crossover since 2023. New Fed chair Warsh inherits a market pricing hikes while his instinct is cuts. 10-year at 4.46%. Cheap money is not arriving on any schedule that helps 2026 planning. Platform-layer players consolidating while application-layer gets squeezed.
The Compute-Scale Thesis Has a Price Tag — And Chinese Labs Just Published It
The Numbers That Rewrite the Budget
Data from fourteen Chinese AI labs undercuts the premise behind roughly a trillion dollars of planned US AI infrastructure spending. The labs are extracting 4-7x more performance per compute unit than US scaling curves predict. The deployable compute gap widened from 3x in 2023 to 8x in 2026, and yet the capability gap is six to eight months. Those three numbers do not sit comfortably together unless the scaling curves are wrong.
The constraint created the capability. Export controls forced efficiency innovation that is now a permanent Chinese advantage. Removing the controls does not remove the capability.
The Pricing Floor US Labs Cannot Match
DeepSeek V4 Pro prices at $0.43 per million input tokens against roughly $4.73 for Claude Opus, an 11x gap on input and 28x on output. Z.ai ships GLM-5 at a dollar per million tokens and reports 50% gross margin. MiniMax claims 70% enterprise margins at comparable prices. A reasonable skeptic would call these loss leaders. The margin disclosures say otherwise, and the cost to serve is genuinely lower.
China's token volume now runs 2.25x US and Western volumes combined. Nine quadrillion tokens per month against four. That is an inference-stage flywheel, and it compounds the efficiency advantage rather than sitting beside it.
Why This Isn't Just a China Story
The same economics are showing up domestically from a different direction. Research has now produced 4-billion-parameter recursive language models matching Claude Sonnet 4.6 through architecture rather than scale. Cactus Needle distilled Gemini into 26M parameters running at 6,000 tokens per second on consumer hardware. Perceptron entered video analysis at 80-90% below incumbent pricing.
Two other angles arrive at the same destination. Academic research is producing small-model parity without scale, and AI-native businesses are landing at 17% gross on the P&L side. Add the Chinese efficiency story and three independent pressures agree on one claim: the premium attached to frontier access has a shorter shelf life than the contracts being signed against it.
What This Means for the AI P&L
Run the math at 1 million users and $120 ARPU under current inference pricing. The result is 17% gross margin, 11% net. Reasoning models consume 10 to 100x more tokens than their predecessors, which offsets the per-token price drops entirely. The five-year LTV projection built on 80% gross margins describes a business that does not exist at this cost structure.
The Strategic Fork
Jensen Huang joining Trump's China delegation is the tell that export controls are being renegotiated. If restrictions ease, Chinese labs pair frontier silicon with efficiency work they have already done. The compute-constrained-China premise has a shorter half-life than most 2026 plans assume.
The decision this quarter is whether the AI spend assumes a durable efficiency premium on frontier hardware, or assumes that premium compresses as Chinese techniques diffuse. The two assumptions produce very different twelve-month P&Ls. Most capex plans quietly assume the first. The second is now the more defensible forecast, and the gap between defensible and quietly assumed is where next year's write-down lives.
- Update: Shai-Hulud supply chain worm now wipes systems when operators attempt token revocation — creating a hostage dynamic that inverts standard incident response playbooks
- Update: Microsoft -16% YTD (worst big tech performer) as TCI exits — market actively discounting conglomerate structures against AI-focused purity plays
- Employees at Amazon, Meta, and Microsoft are gaming internal AI adoption metrics — running unnecessary tasks to hit leaderboards, meaning your own AI KPIs are likely measuring performance theater, not value
- April CPI hit 3.8% against 3.6% wage growth — first negative real-wage crossover since 2023, with new Fed chair Warsh inheriting a market pricing hikes against his dovish instincts
- shadcn/ui has become the de facto standard for AI-generated interfaces across Figma Make, Cursor, and Claude — your design system is being routed around one PR at a time
- Coinbase x402 processed 178.7M agentic payments ($42.4M) through Amazon Bedrock — stablecoin rails settling in fractions of a cent vs. card networks' 3-4 cents, with no viable alternative at scale
- 21% of top ML conference peer reviews are now fully AI-generated — detection works against bulk slop but skilled writers evade trivially, making provenance infrastructure the only durable answer
- Jensen Huang boarded Air Force One for Trump's China trip — Nvidia's presence signals export control regime is being actively renegotiated, not defended
The AI cost structure just inverted from three directions simultaneously: Chinese labs deliver comparable capability at 4-7x the compute efficiency and 10-28x lower pricing, small models match frontier performance at 4 billion parameters, and AI-native businesses structurally cap at 17% gross margins instead of SaaS's 70%. Any AI strategy, budget, or vendor contract built on frontier-pricing assumptions has a 6-12 month shelf life — and the companies gaming their own AI adoption metrics internally are proving that most organizations haven't even honestly measured what they're getting for the spend they've already committed.