Redpoint Ventures just put out their latest 70-slide market update this month that is one of the most data-dense look at where B2B software actually stands right now. I read the whole thing. Here are the 10 learnings that matter most for B2B + AI founders and operators.
1. Public SaaS Multiples Are at 4.1x NTM Revenue — the Lowest Since 2008
The median NTM revenue multiple for public SaaS companies is currently 4.1x. The peak in 2021 was around 22x. That is an 80% contraction in what the market is willing to pay for a dollar of forward software revenue.
What is driving this is not current-quarter results. The Redpoint data makes a sharp point here: in a typical SaaS DCF, roughly 85 to 95% of enterprise value comes from terminal value, not near-term cash flows. Investors are not debating this quarter. They are pricing in whether AI compresses long-term growth rates, moats, and addressable markets permanently.
The implied long-term growth rate baked into current valuations for a representative public SaaS company has dropped from 4.7% just three months ago to 1.1% today, with the 5-year forward revenue CAGR assumption unchanged at 12.5%. That is the market pricing in near-zero perpetuity growth.
2. Software Stocks Are Down 20% YTD While Energy Is Up 32%
Year-to-date total returns across S&P 500 sectors show software sitting at negative 20%, making it by far the worst performing sector. Energy is up 32%. Materials up 5%. Consumer staples up 4%. Even IT broadly is only down 11%.
The differentiation within software is stark. Horizontal SaaS is down 35% over the last twelve months. Vertical SaaS is up 3%. Infrastructure is up 2%. The market is expressing a view that horizontal software — tools built to serve every industry equally, which often means deep integration with none — is the most exposed category as AI rewrites the coordination and workflow automation problem.
3. 54% of CIOs Are Actively Pursuing Vendor Consolidation, and 45% of AI Budgets Are Replacing Existing Software Spend
Redpoint surveyed 141 CIOs in March 2026. The numbers are not friendly to incumbents. This may be the most important learning for founders & B2B execs.
- 54% are actively pursuing vendor consolidation.
- 45% say their AI budgets are coming directly out of existing software line items, not new budget.
- Only 3% expect AI to lead to more vendors overall.
- 58% say AI feature additions are the number one driver of software spend increases — the highest of any category surveyed.
The practical implication is that AI spending is largely zero-sum for the existing software stack. When a company buys an AI tool, it is often canceling or reducing something else. Net new ARR acquisition is harder because AI-native startups are causing buyers to pause before committing to new seats in incumbent platforms.
4. 83% of CIOs Are Open to Replacing Their CRM with an AI-Native Vendor
When asked which software categories they are most open to replacing with an AI-centric alternative, CIOs ranked Salesforce Automation first at 83%. Customer service management was second at 56%. ITSM at 55%. ERP and procurement tied at 50%.
The categories at the bottom of the list — finance operations at 14%, DevOps at 19%, project management at 19% — are either deeply embedded in technical workflows or carry too much integration complexity to move quickly.
For AI-native founders, the message is clear: CRM is the most contested category in enterprise software right now, and enterprise buyers are saying so directly.
5. AI-Native Companies Are Generating 10x More Revenue Per Employee Than Legacy Software
ARR per full-time employee at current levels: Cursor is at $6.1M per FTE. Lovable is at $3.4M. OpenAI is at $1.5M. Anthropic is at $1.2M. Salesforce is at $0.54M. Datadog at $0.51M. ServiceNow at $0.49M. Atlassian at $0.46M.
This is not a marginal efficiency gap. Cursor is generating 12x more revenue per person than Salesforce. This data point has significant implications for how we think about what a B2B software company should look like structurally — headcount, cost structure, and the relationship between revenue and people.
6. Private Market Series B/C Multiples Are at 61x ARR, While Public High-Growth Software Trades at 9.7x
The spread between private and public software valuations is historically extreme. Median Series B and C ARR multiples in 2026 YTD are 61.1x. Public high-growth software (top 20th percentile by growth rate) trades at 9.7x LTM revenue. That is a 528% premium for private over public.
The growth-adjusted story is more nuanced. Private companies at these stages are growing median ARR at 640% year over year. Public high-growth software is growing at 29%. On a growth-adjusted basis, private software multiples are actually at 0.05x versus 0.37x for public — an 86% discount to public markets on that metric.
The market is effectively saying that private AI-native companies at Series B/C deserve massive premium multiples because their growth rates are in a completely different category than anything public markets have seen.
7. Among Companies That Have Reached $50B+ Valuation … the Recent Cohort Got There in a Median of 9 Years vs. 23 Years for Earlier Generations
Pre-2000s cohort: 23-year median to $50B. The 2000s cohort: 16 years. Post-2000s cohort: 9 years. And within that last cohort, Anthropic hit roughly $50B in 4 years. Cursor did it in 4 years. xAI did it in 1 year.
This compression is a function of faster product-market fit cycles, lower distribution costs, and — critically right now — the underlying model capability advancing faster than any previous platform shift. The cloud transition took roughly a decade. The AI application layer is compressing that timeline significantly.
8. The AI Application TAM Is $6.1 Trillion If Agents Move Beyond Task Execution
Redpoint lays out the agent maturity curve across four stages: copilots (seconds of independent runtime), task agents (minutes), workflow agents (hours), and fully autonomous systems (days). The addressable market grows at each stage.
US software spend alone is $0.5T. Add services automation and you reach $1.2T at the task agent layer. Add operational payroll and the figure reaches $2.8T. Add knowledge worker payroll and the total addressable market is $6.1T+. US professional labor payroll alone is over $6T. Even 5% AI penetration of that number — potentially conservative given 85 to 90% cost reduction in early deployments — exceeds the entire existing US software market.
We are still largely at the task agent stage. Workflow agents that run for hours and autonomous systems that run for days are not yet the standard deployment model. That means the vast majority of the total opportunity has not been captured yet.
9. 44% of All Enterprise Software VC Dollars Are Now Concentrated in the Top 20 Deals
In 2020, the top 20 deals represented 8% of all enterprise software VC funding. In 2021, 6%. In 2022, 7%. Then the AI era hit: 2023 jumped to 23%, 2024 to 31%, and 2025 to 44%.
Capital is not spreading democratically across the AI ecosystem. It is concentrating into a handful of names — and looking at what those names are, they are almost entirely foundation model labs and infrastructure plays. OpenAI, Anthropic, and xAI appear in the top 5 deals in 2023, 2024, and 2025 consecutively. For application-layer founders, this creates a real fundraising dynamic: median round sizes and valuations are rising at every stage, but the dollars are piling disproportionately into the very top of the market. Being second in your category is a harder position to fund than it was three years ago.
10. AI Is Creating More Demand for Software Engineers, Not Less
Indeed job postings for software engineers have diverged sharply from total job postings in early 2026. Total job postings on Indeed are tracking around 85 on an indexed basis (down from 100 in January 2024). Software engineer postings are back near 97 — nearly recovered to baseline and rising.
The historical parallel here is instructive. When ATMs were introduced in 1973, the New York Times predicted they would replace up to 75% of bank tellers. Instead, teller employment expanded by 81% from 1970 to 1988. Lower operating cost per branch made it economical to open more branches, which required more tellers. The same dynamic appears to be playing out with software: lower cost of building software is generating demand for more software to be built, which requires more engineers to direct, review, and deploy it. The net headcount effect of AI coding tools may be the opposite of what most people expect.
What This Means for B2B + AI Founders Right Now
The numbers point in one consistent direction. The application layer opportunity is larger than traditional software TAMs by an order of magnitude. But public markets are pricing existing software at terminal-value-near-zero because the disruption risk to incumbents is real.
For founders building now: the categories most exposed are horizontal coordination software, seat-based pricing models, and anything where the core value proposition is workflow management without proprietary data. The categories with structural protection are vertical software with years of embedded industry-specific data and infrastructure where AI workloads create demand rather than displacement.
The 2026 to 2027 window, based on historical patterns across the internet, cloud, and mobile platform shifts, is when durable category winners tend to get established. That is now.
Grab the full report here.
