2026 VC Predictions Market Map
We analyzed what 132 investors expect from 2026 and mapped it into a single framework
The Prediction Cycle
Starting in December 2025 and continuing into early 2026, I began to notice how deeply the market had slipped into what I’d call “prediction mode” – a state where almost everyone felt compelled to publicly articulate their version of the future, even when that future was still highly uncertain.
Investors, founders, operators, journalists – nearly everyone I follow started publishing their own outlooks:
where capital would flow
which domains would overheat
whether liquidity would return
whether the window for mega-IPOs would reopen
and much more
In isolation, none of this was new. The industry has always tried to look ahead. But what felt different this time was how many parallel storylines appeared at once and how poorly they spoke to each other. You could read ten different “2026 outlooks,” nod along to each of them – and still have no real sense of how those views interacted, overlapped, or contradicted one another.
There was no common frame, no shared coordinate system, not even a rough map of which convictions were widely shared and which were held by one or two loud voices. Everything lived in its own micro-context, optimized for distribution rather than synthesis.
This sense of fragmentation is what ultimately led to this Market Map.
I’m not trying to predict the future here. I’m trying to document how the market is thinking about the future right now – with all the contradictions, overlaps, and blind spots that come with that.
How This Map Was Built
Before turning to the insights, it’s important to explain the methodology. This project is primarily about capturing how investors talk about uncertainty and where they are willing to place actual capital behind those stories, rather than validating whether specific forecasts will turn out to be “right.”
Sources
Most of the data was collected from public sources:
Analytical articles and industry roundups
Substack newsletters
LinkedIn
X
Personal blogs
Fund websites
Public interviews
In total, we initially identified close to 300 public sources, including journalists, researchers, venture investors, founders, and operators.
Given the volume and heterogeneity of perspectives, we deliberately narrowed the scope to VC investors, as they are most directly responsible for capital allocation.
Even within this group, many views were fragmented, incomplete, or insufficiently grounded to support reliable interpretation. Some predictions were little more than slogans or marketing copy. Others lacked any link to how a fund actually invests. As a result, we excluded a number of cases that did not meet our internal consistency thresholds.
The final dataset includes 132 venture investors.
Domains
We structured the map around 9 core clusters. Each could be further subdivided, but for this version we intentionally stayed at a higher level of abstraction, because once you slice the world too finely, it starts to look precise on paper while becoming less useful for real decisions.
Agentic Systems
This cluster covers autonomous and semi-autonomous AI systems capable of executing multi-step tasks with minimal human supervision. It includes agent frameworks, orchestration layers, multi-agent collaboration, AI-native workflows, and emerging agent-to-agent commerce and coordination models.
AI Infrastructure
This domain includes the technical backbone of AI development and deployment: compute, chips, cloud optimization, model serving, inference stacks, MLOps, developer tooling, and platforms that reduce the cost and complexity of building and scaling AI systems.
World Models & Planning
We use this category for systems focused on long-horizon reasoning, simulation, and environment modeling. This includes foundation models for planning, reasoning engines, digital twins, reinforcement learning at scale, and architectures designed to build internal representations of the world.
Physical AI & Energy
This cluster spans the intersection of AI with the physical world: robotics, autonomous systems, industrial automation, embodied AI, energy optimization, grid intelligence, and infrastructure required to support compute-intensive workloads.
AI Economics & Pricing
This domain captures business-model innovation around AI: usage-based pricing, outcome-based contracts, agent marketplaces, compute monetization, cost-passing mechanisms, and the emerging economics of deploying AI at scale.
Data & Memory Moats
We group here companies building defensible advantages through proprietary data, long-term memory systems, feedback loops, retrieval architectures, and domain-specific datasets that compound in value over time.
Talent & Distribution
This cluster reflects the human and go-to-market layer of AI: platforms for hiring and managing technical talent, developer communities, AI-native sales and marketing systems, workflow distribution, and mechanisms for acquiring and retaining users in increasingly crowded markets.
Fintech & Financial Rails
This domain includes payment infrastructure, embedded finance, compliance tooling, agent-native wallets, settlement systems, cross-border rails, and financial primitives designed to support AI-driven commerce and automation.
Liquidity & IPOs (Market)
Unlike the other clusters, this category captures market-level dynamics rather than technology. It includes expectations around exit environments, public market receptivity, M&A activity, secondary liquidity, and structural shifts in how venture-backed companies achieve liquidity.
Marker System
Each cell includes markers designed to clarify what an individual meant – especially in cases where their original statements were compressed, implicit, or spread across multiple posts and interviews:
● Full circle – strong conviction
◐ Half-filled – tentative hypothesis
Empty – topic not addressed or deprioritized
Whenever possible, we added short contextual notes. Not to reinterpret statements, but to preserve their original intent while making them legible inside a larger system.
AI Support and Manual Verification
For analysis and clustering, we used:
Perplexity Pro Research
ChatGPT Pro
Grok
Claude
Other research engines
These tools helped us scale the process and detect recurring patterns across large volumes of text that would have been impractical to read line by line.
At the same time, interpretation remained human. We manually reviewed every participant, every source, and every extracted prediction, because I’ve learned that without this layer of friction, it’s very easy to mistake surface-level similarity for genuine alignment.
Top-5 Bets
Based on aggregated signals, we identified 5 themes that appeared most frequently across predictions. These are not “best ideas.” They reflect collective market attention.
#1: Autonomous AI as the Next Platform Layer
Agentic systems are the most frequently cited theme across the dataset.
Investors are increasingly focused on AI systems that can operate autonomously, coordinate with other agents, execute multi-step workflows, and interact directly with digital and financial infrastructure.
This goes far beyond chatbots or copilots. What many investors are underwriting is the emergence of:
multi-agent workflows
agent-native products
agent-to-agent commerce
automated business processes
Underneath this, there is a simple belief: that productivity gains from AI will move from “human + AI” to “AI + AI,” with humans supervising rather than operating. In other words, most of the work will happen between systems in the background, not in a visible UI between a person and a chatbot.
I’m not sure this is the future everyone wants, but it is the one I see most often between the lines of these predictions. In 2026, early signs of this shift are likely to show up in:
early agent marketplaces
primitive autonomous companies
vertical-specific agents
growing tension around reliability and governance
This is the closest thing in the dataset to a “new platform shift.”
#2: AI Infrastructure
The second strongest theme is infrastructure.
This includes compute, inference optimization, deployment stacks, model serving, developer tooling, and cost-efficiency platforms. After two years of foundation-model hype, many investors now see infrastructure as the real bottleneck.
Capital is flowing into companies that help:
reduce inference costs
optimize GPU utilization
manage AI workflows
make AI systems production-grade
This looks less like a new hype cycle and more like a clean-up phase. After two years of foundation-model excitement, investors are trying to fund the “picks and shovels” that make AI cheaper, more reliable, and easier to operate at scale. Over the next year, I expect to see:
consolidation in tooling
pressure on margins
verticalized infrastructure stacks
and fewer “general-purpose” platforms
#3: AI Meets the Real World
This cluster reflects growing conviction that the next wave of AI value creation will happen outside pure software.
Robotics, automation, manufacturing, logistics, energy optimization, and embodied AI appear consistently across predictions.
Two forces are converging here:
AI models are becoming capable enough to control physical systems
energy and compute constraints are becoming strategic issues
Investors are positioning around:
warehouse robotics
autonomous systems
industrial AI
energy management
grid intelligence
This also reads like a hedge. If consumer AI saturates, physical systems offer a longer runway and deeper moats. It is much harder to copy a robotics stack deployed across dozens of factories than a new interface on top of a public API. In 2026, expect:
slower adoption cycles
capital-intensive scaling
but potentially more defensible businesses
#4: Rebuilding Payments for an AI-Native Economy
Fintech reappears in this dataset, but in a very different form than in 2020–2021. The focus is not on neobanks or consumer apps, but instead on infrastructure for AI-driven commerce. This includes:
agent-native wallets
automated payments
compliance tooling
settlement rails
cross-border systems
Many investors are betting that if agents transact with each other, someone must own the rails. I think this is one of the most underappreciated themes. Every time software rewires a workflow, someone quietly rebuilds the payment layer underneath it – and those rails compound for years. AI agents will not be an exception. Key 2026 bets:
early agent payment protocols
regulatory friction
fragmentation before consolidation
#5: AI Economics & Pricing
The fifth major theme is AI monetization. After years of subsidized usage, investors are increasingly focused on how AI businesses will actually make money. Recurring topics include:
usage-based pricing
outcome-based contracts
cost pass-through
compute-indexed fees
margin compression
Many investors explicitly question whether current AI pricing models are sustainable. I think this reflects a shift from “growth at all costs” to “unit economics matter again.” In 2026, expect:
pricing experimentation
customer pushback
consolidation of weaker models
clearer segmentation between premium and commodity AI
Top-5 Selected Quotes
In parallel with mapping dominant themes, we also looked for statements that reflect deeper structural thinking about where the industry is heading. Out of hundreds of public predictions, we selected 5 quotes that we subjectively consider the most fundamental.
#1: David Cahn from Sequoia Capital
Why this matters: This is a quiet but important reframing. When AI becomes infrastructure rather than a product category, the category "AI company" disappears. Only companies that embedded AI into their operational DNA survive. This is the maturation moment where AI stops being a differentiator and becomes table stakes.
#2: Tomasz Tunguz from Theory Ventures
Why this matters: This is an inflection point in economics: when a company pays more for an agent than a human, it means humans no longer compete on price. They must compete on creativity, judgment, and strategic thinking. This fundamentally reshapes the labor market and creates new skill hierarchies.
#3: Macy Mills from A16Z Speedrun
Why this matters: This inverts the entire SaaS pricing model. Instead of paying for access to software, you pay for outcomes. This means AI vendors must be confident their agents will deliver results – and it eliminates the inefficiency of buyers paying for tools they don't fully utilize. This is the shift from seat-based to performance-based economics.
#4: Jeff Morris Jr. from Chapter One
Why this matters: Memory is the missing ingredient that will unlock the next wave of AI products. Just as location-awareness created entirely new categories that didn't exist before, persistent memory will enable products impossible today. The first companies to crack memory-driven products will have defensible moats that last for decades.
#5: Olivia Moore from A16Z
Why this matters: Voice agents are currently tactical tools, but in 2026 they become strategic infrastructure. The shift from "voice for scheduling" to "voice managing the entire customer lifecycle" captures massive value. Companies that expand voice from wedge to platform will generate revenue while saving costs simultaneously.
Conclusion
This is the first iteration of the 2026 VC Predictions Market Map.
It is incomplete by design, and we expect to keep editing and refining it as the year unfolds and today’s narratives either compound into real companies, rounds, and liquidity events (or quietly fade).
Toward the end of 2026, we perhaps plan to revisit this map, review which bets played out and which didn’t, and layer a second pass of analysis on top.
If this work is useful to you, follow Murph Capital and stay tuned. We are preparing the launch of our open-source “new media“ and community, along with a growing set of tools and resources designed to help emerging managers build better funds and LPs make better long-term investment decisions.
good article learn alot from them