BHI Rankings Q1 2026

From commodity chatbots to gravitational wells. The AI systems reshaping how work gets done.

Microsoft Copilot outscores ChatGPT by 2.4x despite NPS=-19.8 and only 8% voluntary adoption. Claude has the highest action capability but the lowest lock-in by design. The AI lock-in race is not about who makes the best model. It is about who embeds deepest into organizational infrastructure.

Top 10 crypto assets by market cap. Not all market cap is backed by structural lock-in.

The deepest crypto lock-in is not in the newest tokens but in infrastructure layers. Ethereum developer ecosystem, USDT payment rail dominance, and BNB exchange infrastructure create structural capture that market cap rankings alone completely miss. Dogecoin at $15B and near-zero B proves that massive capitalization can exist with no structural floor.

Exchanges, DeFi protocols, wallets. Where crypto lock-in actually lives.

Crypto infrastructure lock-in follows a clear hierarchy: oracle standards (Chainlink) and exchange infrastructure (Binance) at the top, lending protocols (Aave) in the middle, and NFT marketplaces (OpenSea) eroding at the bottom. The fork myth persists in DeFi — but liquidity, integrations, and oracle dependencies cannot be forked.

3 billion users mean nothing without a trapped social graph.

WeChat represents the theoretical maximum (p=10, h=1). WhatsApp achieves the highest escape cost through messaging ubiquity. TikTok proves 1.56B MAU with 1hr 35min daily means nothing structurally when the algorithm, not the graph, drives retention. X demonstrates active erosion: rising substitutability from Threads and Bluesky is measurably decreasing B quarter over quarter.

The chokepoints of the global digital economy. From chips to clouds.

TSMC (p=10, s=1) and ASML (p=10, s=1, h=1) represent the two deepest structural lock-in positions in the global economy. NVIDIA CUDA has locked millions of developers with no viable migration path. These three companies form a chain: ASML makes the lithography, TSMC uses it to make chips, NVIDIA designs those chips to run CUDA. Break any link and AI development stops globally.

Enterprise infrastructure. Where organizational lock-in reaches maximum depth.

AWS (p=10) and Salesforce (o=9) represent the extremes of cloud lock-in: infrastructure dependency vs organizational embedding. Zoom proves that massive daily usage with high substitutability produces commodity economics. Snowflake demonstrates data gravity — petabytes of accumulated data are the ultimate switching cost barrier.

The original lock-in machines. Decades of accumulated dependency.

Bloomberg Terminal is the deepest structural lock-in in our entire dataset outside TSMC and ASML. SWIFT (p=10) demonstrates that cooperative infrastructure can create even deeper lock-in than corporate platforms. JPMorgan and Goldman prove that decades-long banking relationships create switching costs that no fintech app can overcome through better UX alone.

Network rails to checkout buttons. Lock-in hierarchy from maximum to commodity.

Fintech lock-in hierarchy is precise: network rails (Visa/MC) > developer infra (Stripe) > primary banking (Nubank) > merchant POS (Block) > commodity checkout (PayPal). PayPal's decline illustrates structural erosion when substitutability rises. Nubank's $68B is justified by deepest personal financial lock-in: being someone's only bank, only credit card, only savings account.

Your Steam library is worth more than your Netflix subscription. Owned assets vs rented access.

Steam's non-transferable game library is the strongest consumer lock-in in digital entertainment. Netflix (325M subs, B<1.0) proves that massive scale with near-zero switching costs produces fragile revenue. The owned-vs-rented divide is the key structural difference: Steam users accumulate assets that deepen lock-in. Netflix users rent access that evaporates on cancellation.

Logistics infrastructure is the ultimate moat. Not who has the most sellers — who delivers fastest.

Logistics is the ultimate e-commerce moat. Amazon (FBA, 91% renewal), MercadoLibre (own cargo fleet, $278B payments), Coupang (93% renewal) all prove: control of physical delivery creates lock-in that pure marketplace network effects cannot match. PDD at $148B and $786B GMV but moderate B demonstrates that price-driven scale without structural depth is inherently fragile.

Where lock-in is measured in patents, protocols, and irreplaceable scientific infrastructure.

Pharma lock-in is invisible to consumer analysis but extreme at institutional level. Illumina controls 70-80% of sequencing — the entire field of genomics is structurally dependent on one company's data formats. IQVIA is healthcare's Bloomberg: decades of patient data that cannot be replicated. Veeva is pharma's Salesforce: compliance amplifies switching costs by 3-5x versus generic SaaS.

The chokepoints beneath the chokepoints. Where failure cascades globally.

Synopsys/Cadence (EDA duopoly) is the most invisible and arguably deepest lock-in in the global economy. Without their software, TSMC cannot receive chip designs. Without TSMC, NVIDIA has no chips. Without NVIDIA, AI stops. The dependency chain: EDA → Foundry → Chip Designer → AI Company → End User. Break any link and the cascade is global.