Companion view
A living taxonomy for AI market boundaries, leaders, metrics, and model cards.
Cloud had IaaS, PaaS, and SaaS. AI needs a different layer cake: infrastructure, model portfolios, raw data management, AI-ready data, developer tools, agent tools, and commercial tools. The strategic fight is whether SaaS keeps owning the data, or whether enterprises own their data and expose it safely to agents.
7
Categories
50
Seeded players
19
Leaders
21
Metrics
Reference architecture
The new boundary is data access. Models matter, but the market structure forms around who owns the data, who makes it AI-ready, who builds with it, which agents can act on it, and which commercial tools survive on top.
L1 Infrastructure
Physical and cloud substrate for AI: GPUs, accelerators, networking, datacenters, power, storage, and managed compute capacity.
Leaders
Amazon Web ServicesControl points
compute / power / networking / cloud_capacity
Watch
L2 Models
Foundation, frontier, open-weight, specialist, and routed model portfolios that provide the intelligence primitives consumed by higher layers.
Leaders
Amazon Web ServicesControl points
weights / benchmarks / token_pricing / model_policy / routing
Watch
L3 Data Mgmt
Raw systems of record, warehouses, lakehouses, operational databases, and document stores that hold enterprise data before it is shaped for AI use.
Leaders
Control points
source_data / governance / lineage / access_control
Watch
L4 AI Data
Processed, indexed, embedded, retrievable, policy-wrapped, and API-exposed data that can be consumed directly by models and agents.
Leaders
Amazon Web ServicesControl points
embeddings / vector_index / retrieval / api_gateway / context_layer
Watch
L5 Dev Tools
Build tools for humans and agents creating software: IDE copilots, coding agents, CLI agents, context management, orchestration, evals, and tool harnesses.
Leaders
Amazon Web ServicesControl points
developer_workflow / context_management / orchestration / tool_harness / evals
Watch
L6 Agents
Agent runtimes and operational tools that run scheduled, triggered, multi-agent, tool-using, and exception-escalating work.
Leaders
Control points
tools / memory / permissions / task_execution / human_review
Watch
L7 Commercial
Finished point solutions and vertical AI tools: video, voice, legal, sales, research, support, writing, and other specialist business outcomes.
Leaders
Control points
workflow_distribution / proprietary_context / seats / vertical_expertise
Watch
Boundary rule
Vendors get one primary layer by market role, plus secondary roles when they span the layer cake. Confidence is explicit so category ambiguity does not masquerade as precision.
Confidence distribution
19 high, 27 medium, 4 low-confidence placements.
Low confidence means useful signal, not settled category leadership.
Market scorecard
Each layer gets metrics that can be revalidated from public sources. When the data is sparse, the chart says so instead of inventing a false leaderboard.
L1 Infrastructure
3 tracked signals
Q1 2025
90d cadence
L2 Models
3 tracked signals
2024 vs 2023
vs 4.4%
180d cadence
L3 Data Mgmt
3 tracked signals
2024 vs 2023
vs 55%
365d cadence
L4 AI Data
3 tracked signals
current
90d cadence
L5 Dev Tools
3 tracked signals
2025 estimate
365d cadence
L6 Agents
3 tracked signals
March 2025
180d cadence
L7 Commercial
3 tracked signals
2025 estimate
365d cadence
Layer 1 — AI Infrastructure
Data-center capex growth: 53% YoY
AI infrastructure remains capital constrained; capex direction is the first-order signal for who can supply capacity.
Layer 2 — Model Portfolio
SWE-bench performance jump: 71.7%
Software-engineering capability moved from toy benchmark to enterprise relevance, changing the model layer's practical value.
Layer 3 — Data Management
Organizations using AI: 78%
More organizational AI use increases pressure to expose governed systems of record to agents without copying everything into SaaS silos.
Layer 4 — AI-Ready Data
Vector and hybrid retrieval availability: available across Pinecone, Weaviate, Milvus, MongoDB, Elastic
Retrieval infrastructure is now broad enough to treat AI-ready data as a separate layer between raw datastores and agent execution.
Layer 5 — Developer Tools
Coding tools as leading GenAI application category: top enterprise spend category
Developer tools are the first place where AI changes production work, not only information retrieval.
Layer 6 — Agent Tools
Agent startups tracked: 170+
Agentic workflows are a distinct market structure, not just another app feature.
Layer 7 — Commercial Tools
GenAI application spend: $19B+
Commercial tools are already a major value-capture zone, but the durable winners need workflow depth that survives model-provider expansion.
Metrics by category
Each category carries a small metric basket: one or two headline indicators plus supporting signals. The point is not to rank everything every day; it is to keep a durable public watchlist with explicit confidence and refresh cadence.
L1 Infrastructure
3 signals
Data-center capex growth
capital / Q1 2025
high / refresh 90d
AI infrastructure remains capital constrained; capex direction is the first-order signal for who can supply capacity.
High-end accelerated servers
mix_shift / 2025 forecast
medium / refresh 90d
Accelerator-heavy systems are no longer a niche line item; they are a large share of the data-center investment stack.
Data-center capex forecast
capital / 2025 forecast
high / refresh 90d
Sustained infrastructure expansion keeps the AI substrate layer strategic, not commodity.
L2 Models
3 signals
SWE-bench performance jump
capability / 2024 vs 2023
prior 4.4%
high / refresh 180d
Software-engineering capability moved from toy benchmark to enterprise relevance, changing the model layer's practical value.
Open-weight vs closed-weight performance gap
capability / 2024 vs prior year
prior 8.04%
high / refresh 180d
Narrowing gaps increase buyer leverage and make openness a strategic dimension, not only a developer preference.
Live model price, latency, speed, and context comparisons
efficiency / live
high / refresh 30d
Model leadership should be benchmark-adjusted by cost and latency, not read from capability leaderboards alone.
L3 Data Mgmt
3 signals
Organizations using AI
adoption / 2024 vs 2023
prior 55%
high / refresh 365d
More organizational AI use increases pressure to expose governed systems of record to agents without copying everything into SaaS silos.
Generative AI in at least one business function
adoption / 2024 vs 2023
prior 33%
high / refresh 365d
Business-function adoption makes source-data access, lineage, and permissions first-class AI architecture concerns.
Enterprise GenAI spend
spend / 2025 estimate
medium / refresh 365d
Enterprise AI spend eventually lands against data platforms, not only model APIs and point tools.
L4 AI Data
3 signals
Vector and hybrid retrieval availability
capability / current
high / refresh 90d
Retrieval infrastructure is now broad enough to treat AI-ready data as a separate layer between raw datastores and agent execution.
Model and context gateway pattern
architecture / 2026
medium / refresh 90d
API gateways, model routers, and context wrappers become the control plane for which agents can consume which data.
Open-model serving demand proxy
adoption / live
medium / refresh 30d
Open ecosystem activity is a maintainable proxy for AI-ready data and serving demand where vendors do not disclose token volume.
L5 Dev Tools
3 signals
Coding tools as leading GenAI application category
spend / 2025 estimate
medium / refresh 365d
Developer tools are the first place where AI changes production work, not only information retrieval.
Coding agents and CLI tools
workflow / 2026
medium / refresh 90d
Build tools are moving from autocomplete to agent supervision, with context management and tool harnesses becoming the real differentiators.
Context management as build substrate
architecture / 2026
medium / refresh 90d
The durable build-tool layer may be less about the IDE and more about how context, tools, repo state, and review loops are packaged for agents.
L6 Agents
3 signals
Agent startups tracked
market_structure / March 2025
medium / refresh 180d
Agentic workflows are a distinct market structure, not just another app feature.
Agent startup funding
funding / 2024
medium / refresh 365d
Funding growth signals that investors view delegated work as a separate value pool.
Public production-readiness disclosure
disclosure / 2026
medium / refresh 90d
Human-in-loop controls, tool permissions, and audit logs should carry as much weight as demos.
L7 Commercial
3 signals
GenAI application spend
spend / 2025 estimate
medium / refresh 365d
Commercial tools are already a major value-capture zone, but the durable winners need workflow depth that survives model-provider expansion.
Microsoft 365 Copilot paid enterprise seats
adoption / April 2026
high / refresh 90d
Seat adoption is the cleanest public signal for commercial AI distribution through incumbent software suites.
Specialist tool pressure
market_structure / 2026
medium / refresh 90d
OpenAI and Anthropic moving into implementation services raises the bar for point solutions: they need proprietary workflow depth, data access, or domain-specific trust.
Monthly position
This is the living part of the taxonomy: a monthly read of who is gaining, holding, or losing position inside each category. Scores are editorial indices, not market share, and they should be recalibrated as better public signals arrive.
L1 Infrastructure
2026-03 / 2026-04 / 2026-05
NVIDIA
leader
96↑
position index
Merchant accelerator leadership remains the infrastructure benchmark; score rises with continued capex pull-through and networking attach.
Microsoft
leader
89↑
position index
Infrastructure position strengthens when Azure capacity and Copilot distribution reinforce each other.
Amazon Web Services
leader
88→
position index
Cloud capacity and procurement breadth keep AWS in the leader band.
AMD
challenger
70↑
position index
Challenger score improves when substitution pressure and cloud availability become more credible.
L2 Models
2026-03 / 2026-04 / 2026-05
OpenAI
leader
94→
position index
Frontier capability plus product habit keeps OpenAI in the leader band; open-weight pressure shows up as efficiency pressure, not displacement.
Anthropic
leader
91↑
position index
Enterprise trust and agentic coding strength keep Anthropic gaining within the model layer.
leader
89↑
position index
Google benefits from simultaneous model, cloud, and consumer distribution signals.
Hugging Face
leader
82↑
position index
Open-model distribution and model-card gravity keep Hugging Face in the platform leader band.
DeepSeek AI
challenger
81↑
position index
DeepSeek remains the open-weight efficiency challenger forcing price and training-capital discipline.
L3 Data Mgmt
2026-03 / 2026-04 / 2026-05
Databricks
leader
88↑
position index
Governed data gravity keeps Databricks central as enterprise AI shifts from experiments to operated systems.
Snowflake
leader
84↑
position index
Snowflake remains a leader when AI starts from governed enterprise data.
L4 AI Data
2026-03 / 2026-04 / 2026-05
Pinecone
leader
84↑
position index
Pinecone anchors the vector-store portion of the AI-ready data layer.
Weaviate
challenger
74↑
position index
Weaviate gains when teams want open, hybrid, and self-controlled retrieval infrastructure.
Vercel
challenger
71↑
position index
Runtime position improves as model routing becomes part of application deployment rather than a separate platform decision.
Cloudflare
challenger
69↑
position index
Edge distribution and security posture keep Cloudflare relevant as inference gets closer to users and policies.
Together AI
challenger
68↑
position index
Open-model serving demand keeps Together AI in the runtime challenger set.
L5 Dev Tools
2026-03 / 2026-04 / 2026-05
Anthropic
leader
85↑
position index
Claude Code makes Anthropic one of the most important movers in build tooling, not only model APIs.
GitHub
leader
84↑
position index
GitHub owns a high-frequency execution surface where agentic behavior can become daily workflow rather than demo.
OpenAI
leader
82↑
position index
Codex and implementation services move OpenAI upward from model portfolio into the developer-tools layer.
Cursor
challenger
77↑
position index
Cursor gains as coding remains the cleanest category for visible AI productivity and agentic workflow adoption.
LangChain
specialist
66↑
position index
Orchestration and observability mindshare rises as multi-model systems become more common.
L6 Agents
2026-03 / 2026-04 / 2026-05
Anthropic
leader
87↑
position index
Claude's coding and tool-use posture keeps Anthropic central to early production agentic workflows.
ServiceNow
leader
80↑
position index
ServiceNow is positioned where delegated action meets governed enterprise process.
Hermes
specialist
70↑
position index
Hermes represents owned-agent operations over personal infrastructure and owned data, with humans paged only on exception.
Sierra
challenger
69↑
position index
Sierra is a useful challenger signal for customer-facing production agents.
Zeroclaw
specialist
66↑
position index
Zeroclaw is tracked as a future-stack signal for lightweight autonomous execution over owner-controlled tools.
L7 Commercial
2026-03 / 2026-04 / 2026-05
Microsoft
leader
92↑
position index
Paid Copilot seats and enterprise distribution make Microsoft the application-layer benchmark.
leader
88↑
position index
Gemini consumer adoption keeps Google in the leader band even when enterprise monetization is harder to isolate.
Salesforce
leader
81↑
position index
Salesforce stays strong where AI rides existing CRM workflow ownership.
Wiz
leader
80↑
position index
Cloud security posture extends naturally into AI posture as models and agents become part of the attack surface.
CrowdStrike
leader
78↑
position index
Security workflow ownership makes CrowdStrike relevant as AI changes detection, response, and governance boundaries.
Shared responsibility
Cloud became governable when buyers understood what the provider owned and what the customer still had to own. AI needs the same boundary language for accuracy, data exposure, actions, observability, and outcomes.
L1 Infrastructure
Provider owns capacity, availability, hardware lifecycle, and physical resilience. Customer owns workload placement, demand forecasting, and utilization risk.
L2 Models
Model provider owns training, model behavior, release cadence, and safety defaults. Customer owns model selection, workload routing, evals, and fallback policy.
L3 Data Mgmt
Data platform owns storage, lineage, access control, and governance surfaces. Customer owns data quality, semantic modeling, and agent-access policy.
L4 AI Data
AI-ready data provider owns retrieval quality, indexing, wrappers, and access surfaces. Customer owns source-data truth, freshness, permissions, and which agents may consume which context.
L5 Dev Tools
Tool provider owns coding UX, context packaging, model/tool invocation, and review affordances. Customer owns repository permissions, acceptance criteria, and production release governance.
L6 Agents
Agent tool owns execution loop, tool-use guardrails, memory, logging, and escalation mechanics. Customer owns delegated authority, permissions, rollback, and exception policy.
L7 Commercial
Tool provider owns workflow UX, packaged domain behavior, and product controls. Customer owns data portability, agent access, workflow redesign, and whether point tools become durable or get absorbed by agent systems.
Controls tracked across layers
Top model cards
The cards link model lineage to market fit: capability signature, operating envelope, drift watch, and trust surface. They are designed to sit on top of the LLM Evolutionary Tree rather than replace it.
reasoning
OpenAI / Frontier closed reasoning default
Best fit
High-stakes reasoning, agentic coding, multimodal work, and workloads where frontier quality outweighs portability.
Trust surface
Strong enterprise posture through hosted controls; portability and inspectability remain closed-model constraints.
reasoning
Anthropic / Frontier enterprise reasoning and coding model
Best fit
Long-form reasoning, coding, tool-use workflows, and enterprise contexts that weight safety and trust posture heavily.
Trust surface
Strong trust-center posture; still requires application-level output review and action controls.
reasoning
Google + DeepMind / Frontier multimodal reasoning model
Best fit
Multimodal and long-context workloads, especially where Google Cloud or Google product distribution is already strategic.
Trust surface
Strong cloud control surface; model behavior remains closed and requires independent application evals.
reasoning
DeepSeek AI / Open-weight reasoning pressure point
Best fit
Reasoning workloads where cost, portability, and self-hosting leverage matter more than first-party product polish.
Trust surface
Openness improves inspection and hosting control; customers inherit more responsibility for safety, deployment, and monitoring.
mixture_of_experts
Meta AI / Open ecosystem frontier family
Best fit
Enterprise and developer contexts that value ecosystem breadth, inspectability, and deployment flexibility.
Trust surface
Strong portability; customer must own hosting, eval, and policy controls unless using a managed provider.
reasoning
Alibaba / Open-weight reasoning and regional sovereignty signal
Best fit
Reasoning workloads where open weights, regional availability, and China-scale ecosystem signals matter.
Trust surface
Openness aids inspection; jurisdiction, hosting, and data governance need explicit buyer review.
mixture_of_experts
Mistral AI / European frontier and sovereignty model
Best fit
European enterprise and sovereignty-sensitive workloads that need strong performance without a single US frontier dependency.
Trust surface
Regional positioning helps governance narratives; individual deployment controls still determine practical risk.
decoder_only
Cohere / Enterprise retrieval and workflow specialist
Best fit
Retrieval-heavy enterprise workflows, knowledge applications, and workloads that value business-context fit over raw frontier rank.
Trust surface
Enterprise positioning is useful, but buyer-side evals should prove retrieval quality and data handling.
reasoning
xAI / Frontier challenger with social distribution
Best fit
Buyers exploring frontier alternatives and consumer/social-context distribution signals.
Trust surface
Closed-provider controls apply; enterprise trust posture should be treated as less proven than the leading incumbents until evidence improves.
mixture_of_experts
OpenAI / Open-weight strategic hedge from a closed-model leader
Best fit
Portability-sensitive teams that want open-weight leverage without leaving the OpenAI model family narrative entirely.
Trust surface
Open weights shift more operational responsibility to the deployer while retaining strategic familiarity with a leading model provider.
Boundaries. Leaders. Evidence.