AI and the future of premium software pricing

We're moving away from charging for *access* to software and toward a model of charging for the *work delivered* by a combination of software and AI agents. Let’s dive into what’s happening and what it means for you ⤵️ 1. The rise of disruptive AI pricing models Tech companies are realizing they can't solely rely on seat-based subscriptions in an age of AI, automation and APIs where value is disconnected with how many people are logging in. Perhaps Salesforce going all-in on Agentforce (and charging $2 per conversation) was the push the industry needed. Each product category has its own flavor of disruptive pricing. - Legal AI products might charge for a demand package generated by AI or an AI-generated summary. - Creator AI products might charge for the content that gets produced such as a video generation or amount of video created. - GTM products might charge for specific tasks completed or workflows executed by the AI. 2. Selling work, not necessarily success As a customer, I wish I only had to pay for software when it delivered results. But the reality is that true success-based billing won’t work for the vast majority of today’s products. Most products should charge for work output instead. The issue is attribution. You want the customer to get a fantastic outcome — and you want them to recognize that your product powered that outcome. As soon as you start charging for success, the customer begins to rethink the results. 3. Goodbye ARR as we know it? Shifting to these newer value-based pricing models isn't a simple pricing change you can just announce in a press release. It's a business model evolution that looks a lot like the shift from on-prem to SaaS in the first place. These new AI pricing models might mean greater volatility in both usage and spend. Variable margin profiles across products and customers. Seasonal revenue fluctuations. The potential for project-based, non-recurring use cases. Put simply, annual recurring revenue (ARR) continues to get dethroned. — Full post in today’s Growth Unhinged newsletter: https://lnkd.in/ea5eTrVD Things are about to get interesting 🍿 #ai #pricing #saas

Salesforce just fired the starting gun on a seismic shift in how we pay for software. At Salesforce #Agentforce, they announced they’re moving away from the traditional per-seat SaaS model to a consumption-based pricing for their AI agents. This is huge. Why? Because it signals the end of paying just to have access to technology. Instead, we’re moving toward paying for outcomes—the actual value delivered. Think about it. In a world where AI agents can perform the job functions of entire departments, does it make sense to charge per seat? Probably not. Here’s what’s changing: - From access to outcomes: Companies will pay for what the AI actually accomplishes. - From subscriptions to value: Pricing adjusts based on usage and results. - From Software-as-a-Service to Agent-as-a-Service: Technology that collaborates with you as a partner This isn’t just a tweak in pricing—it’s a radical upending of commercial models for large SaaS companies. What does this mean for businesses? - Budgeting will evolve: Costs align directly with value received. - ROI becomes clearer: Easier to measure the direct impact of technology investments. - Greater flexibility: Scale usage up or down based on needs without worrying about seat counts. It’s an exciting time, but also a challenging one. Is every SaaS company ready to embrace a model where companies pay directly for the value they receive? At Uniti AI, we’ve been thinking along these lines. We price our AI agents based on the amount of work they do, not on how many seats a company has. I believe this is the future. What do you think? Is the per-seat model on its way out?

25% of B2B companies expect to use outcome-based pricing by 2028. That's a 5x increase from today's 5%, according to Kyle Poyar's latest research.   This will be a painful, but ultimately healthy transition.   Buyers never wanted software in the first place. They wanted solutions. As AI handles more work end-to-end, pricing migrates from inputs (seats, tokens, usage) to outcomes (cases closed, revenue recovered, risk reduced). Less "how much did you use?" More "did it actually work?"   Thought experiment: if code becomes a commodity and features ship instantly, value shifts from building features to guaranteeing execution. You’re not selling software—you’re selling outcome insurance.   Objections are real—attribution is messy, procurement habits are sticky, and buyers hate surprises. But these are solvable with instrumentation, shared definitions of success, and clear guardrails (Manny Medina). Over time, buyers will demand outcome-based pricing because it reduces their risk. Where outcome-based pricing already fits well: AI-enabled services. Services own end-to-end execution, so attribution is clean and incentives align. Mechanical Orchard is a great example—using AI to move mainframe workloads to the cloud, taking ownership of the entire journey. When you own the “last mile,” charging for success becomes straightforward.   AI customer support vendors have also been pioneers of this model. More vendor types are on the horizon.   If you’re a founder, here’s a simple path to test outcomes pricing: • Pick one mission-critical outcome your product directly influences. • Define a verifiable metric, baseline, and observation window with the buyer. • Cap downside (floor) and share upside (tiers/bonus) to build trust. • Instrument attribution now—event logs, holdouts, and third-party validation beat hand-waving later.   Start with one outcome. One customer. One measurable result you can guarantee. We're still early in this shift, but the direction is clear.   For those already experimenting with outcome-based pricing, what's been your biggest surprise? And for those that haven't yet, what's holding you back?

Jevon & Veblen walk into a data center. The dominant motif around AI has been Jevon’s Paradox : the cheaper a product becomes, the more it is consumed. Token prices dropped 10-20x over the past 18 months & demand exploded in response. Anthropic surged past $19 billion in run-rate last month, up from $9 billion at the end of 2025. OpenAI topped $25 billion in annualized revenue in February, a 17% increase in two months. We know GPUs, CPUs, & memory are already in short supply. Rumors of next-generation models, including Claude Mythos, suggest pricing that moves in the opposite direction. This weekend, an accidental data leak revealed Anthropic’s secretive Mythos model. A leaked blog post described it as : “A step change” in capability, “dramatically higher scores on tests of software coding, academic reasoning, and cybersecurity.” Anthropic stated the model is “very expensive to serve & will be very expensive for customers.” Some have speculated on inference pricing 5-6x more than existing models. If these rumors hold, the most powerful intelligence would trade at a stiff premium. Jevon’s Paradox would give way to Veblen goods. Veblen goods are those whose demand increases with price : front-row concert tickets that cost 10x more despite worse acoustics. Nike Jordans that retail for $110 and resell for $500+. Ivy League tuition where selectivity is the value proposition. Could AI follow this dynamic for competitive advantage? The company with capital to access the most powerful model wins. How much is that worth? Consider a Series A founder building an AI coding assistant. Today, she pays $25 per million output tokens for Opus 4.6. Her burn rate assumes that price. If Mythos launches at $150 per million tokens, 6x more, she faces a choice : raise prices, raise capital, or watch her AI-native competitor ship features she can’t match. The token-maxxing era ends. Companies will stop optimizing for cheap inference. They’ll deploy capital aggressively, both GPUs & dollars, to maximize capability rather than minimize cost. Balance sheets become a moat. The most profitable companies or those who can raise capital cheaply will have the biggest advantage in their industries. For companies that cannot respond quickly enough or afford the most sophisticated AI, the gap widens. If AI-native companies can build 10x faster with Mythos-class models while competitors are stuck on Opus 4.6, valuations will diverge further. Jevon & Veblen walked into a data center. We don’t yet know who walks out.

Sorry to break an uncomfortable truth to SaaS founders Your fancy new AI features? They're eating your margins for breakfast. And lunch. And dinner. The math is brutal but simple: In the good old days of traditional SaaS, for every $100 your customer paid, you kept $80. Your costs were predictable - some servers here, a support team there. Life was good! Enter 2025 and the AI revolution. Now for every $100, you're lucky to keep $65. Why? Because every time your customer uses your shiny new AI feature, ka-ching! The meter's running. And those GPT-4 tokens? They add up faster than a Tesla's battery drains in winter. "But wait!" you say, "We'll just pass the costs to our customers!" Sorry, friend. Your customers aren't stupid. They can do math too. Here's what's actually happening: - Your competitors are absorbing the AI costs to stay competitive - Your customers are expecting AI features as standard - Your margins are getting squeezed from both ends - And your investors are still expecting that magical 80% number 2025 is the year SaaS companies face a choice: Accept that the 80% margin party is over, or completely reinvent how they charge for value. Want to know who's going to win? It won't be the companies with the fanciest AI. It won't be the ones with the lowest prices. It'll be the ones who figure out how to charge for outcomes instead of subscriptions. The future belongs to those who can answer one simple question: What are your customers actually paying for - your software, or the results it delivers? Time to get creative, folks. The margin party's over and the innovation game is just beginning. 🎯 Who's ready for this conversation?

As software evolves from supporting work to performing work itself, the market opportunity can expand beyond traditional software budgets and into labor and services spend. Bain estimates AI-enabled software could create between $5 and $7 trillion in value by 2030 but we believe capturing that value will depend on more than product adoption alone. One of the most important factors shaping AI economics is inference cost: the cost incurred every time an AI system generates a response, completes a task or acts on behalf of a user. As AI usage scales across enterprises, those costs can increase significantly and place pressure on software margins if not carefully managed. This is already becoming a real operational and financial consideration for companies deploying AI at scale. Vista Equity Partners' Monti Saroya and Lily Perez outline a practical framework for understanding how infrastructure, model selection and workflow design can influence both performance and long-term economics in AI-enabled software. https://bit.ly/3PRL4H4

$7,225 for one day of coding. And Cursor isn't even the worst example. Replit's margins went negative. Anthropic throttles its best users. I mapped pricing across 50 AI startups. Six distinct patterns emerged. The core tension: traditional SaaS has near-zero marginal cost per user. AI products pay for compute on every interaction. A casual Claude user costs pennies. A developer running Claude Code all day costs tens of thousands per month. Your best users are your most expensive users. That tension is breaking every pricing model in the market. Cursor charged a flat 500 requests/month. Worked fine until users leaned into multi-step agent workflows. They switched to credit pools. One developer burned 500 requests in a single day. The plan description changed from "Unlimited" to "Extended" twelve days after launch. Replit grew 15x in ten months ($16M to $252M ARR). But they were buying revenue with compute. When they launched a more autonomous agent, margins crashed to negative 14%. They had to invent "effort-based pricing" mid-flight. Anthropic played it differently. Their $17/$100/$200 tiers map to genuinely different user personas, not volume bands. A casual user and a Claude Code developer are different products with different willingness to pay. The lesson across all 50 companies: before you set any price, pull the cost distribution. What does your P10 user cost? P50? P90? If the ratio exceeds 10x, flat pricing will break. In AI products, it almost always exceeds 10x. Full guide with all 6 models, 4 case studies, and a decision tree: https://lnkd.in/gdKaQSMk

Q1 2026 shut down the AI hype debate. AI has shifted from a productivity layer to end-to-end execution, making entire software categories economically unviable. Anthropic hit a $19B run rate, Cursor doubled to $2B, and Lovable added $100M in a month. The scale of the shift is hard to ignore. Meta, Alphabet, and other tech giants plan to invest $650B in AI this year, roughly 3x what they spent 24 months ago, more than the U.S. Interstate system adjusted for inflation. But the most important change was not financial. It was structural. This quarter made one thing clear. The advantage is no longer about who has the biggest or most advanced general-purpose model. It's about who can put models to work inside real workflows. In March, Intercom, Cursor, and Decagon showed the same pattern: domain-specific models trained on proprietary data outperform general models on cost, speed and accuracy, with Decagon running over 80% of its workload on its models. The prevailing concern in 2025 was whether AI would live up to the hype. That concern disappeared this quarter, replaced by a harder question. What happens when an AI agent can do what a $50 per seat SaaS product does? When software shifts from licensed seats to executed outcomes, the economics change quickly. AI stops being a feature layered on top of software and starts replacing entire layers of it. That dynamic sits behind what some are calling the SaaS Apocalypse. The language is dramatic. The math is not. Every week, I sit with CEOs, CIOs, and CTOs in healthcare, financial services, energy and sovereign industries. The pattern is consistent. Access to AI is no longer the constraint. Putting it into production, securely, reliably, governed and at scale, is. That is the work we do at Rackspace Technology. We operate the infrastructure, the data layer and the software integration as one system so AI runs in production, not in pilots. The leadership implications are already evident. Seat-based software becomes a cost to challenge, not a given. Pick one workflow you pay for per user today and test replacing it with an AI agent end-to-end. Not assist. Replace. Proprietary data becomes a competitive moat again. Identify two or three datasets that would materially improve model performance and lock them down now. Governance becomes strategy. Inventory every model or agent in your environment. Who owns it? What data does it touch? What decisions can it make? How can it be shut off? SaaS renewal cycles become strategic moments. Review your next five renewals and ask a hard question. If an agent can do this job in 12 months, should this product still exist? The question is no longer whether to adopt AI. The question now is whether your organization will lead this transition or trail behind it. The gap widened meaningfully in Q1. Most companies will not lose because they ignored AI. They will lose because they assumed the old software model would survive it. Q1 made that assumption expensive.

AI is getting cheaper. Over the past 12–18 months, more operators and investors have begun saying the same thing: AI is breaking SaaS economics. Marginal cost is back. Seat pricing is under pressure. Inference is not free. All of that is directionally correct. But it is incomplete. The more important question is not whether AI introduces marginal cost. The question is who captures the efficiency gains as inference gets cheaper. Some products expand margin. Some expand revenue. Some quietly subsidize heavy users. Others compress under competition. The difference is not model quality; rather, it is cost exposure and pricing architecture. I have been developing a framework to think about this, which is what I call the Jevons Capture 2×2. It maps products (not companies) across two structural dimensions: who bears inference cost and volatility, and who retains efficiency gains as AI becomes more efficient. From that lens, four economic states emerge: Structural Winners, Defensive Beneficiaries, Margin Tension, and Consumer Surplus Trap. More importantly, products migrate between them. Bundled seats can move to credits or metering. Bounded optimization surfaces can expand into open-ended agentic execution. Structural Winners can compress if competition erodes pricing power. Much of the public discourse stops at “move to usage-based pricing.” This piece connects Jevons effects, unit economics and product-level P&L mechanics...AND introduces a migration framework to think about how these states evolve. The theme may now be common. The framework is not.

AI Isn't the Bubble, It's the Wrecking Ball for Software Valuations Software is not going anywhere, it is here to stay, yet six weeks into the new year, the S&P Software Index is down nearly 25% y-t-d, while the S&P 500 is flat. Last year, the dominant AI narrative was “Is AI a bubble?" Well, “AI is not a bubble,” so that narrative will likely prove to be false. Today, many surmise that software will become obsolete; however, that will prove to be false too. Valuations for software companies have undergone a dynamic and profound change that truly refines how we should view the advent of AI. For credit investors: - High Yield bonds are relatively insulated from valuation adjustment since HY Index is comprised of just 3% software. - BSL market has 13% exposure to software; however, first-lien debt typically provides 5-6x debt-to-EBITDA so when valuations collapse to 15x in public market, 10x in private market, most software companies will be able to find a solution to refinance at 4x- 5x debt-to-EBITDA which is the appropriate level. - Direct Lending Funds and BDC extended too much leverage with 23% exposure to software sector, providing financing at 9-10x debt-to-EBITDA. True leverage is higher properly correcting EBITDA adjustments to actuals versus pro-forma. This valuation re-rating does not signal software’s obsolescence; the industry will endure, however, there are a few key considerations when considering the path ahead: - Public software P/E multiples have fallen from ~30x to ~15x (Private companies should be lower). The P/E multiple compression reflects the market refusing to pay 30 years of earnings when AI erodes moats, shortening perceived durability to ~15 years. - The ARR model faces fresh scrutiny; AI-powered SaaS pricing likely shifts to a token-consumption billing model. Token/usage-based pricing has becomes AI-native SaaS pricing model (70%+ adoption in enterprise contracts), which will force many incumbents into hybrid tiers, and sparking a wave of M&A as AI winners consolidate niches. - The most leveraged software companies, those acquired by Private Equity at 18x–20x, rely heavily on direct lending because the broadly syndicated loan market tops out at ~6x. Software companies fall into three cohorts: 1. AI winners that embrace the technology, capture market share, and see soaring enterprise value. In this case, the loan is repaid at par, and equity reaps windfall profits. 2. Software survivors with unique niches that gain efficiency and cost advantages through AI adoption. More supportive for credit vs. equity. Not an earnings problem, but a valuation problem. 3. Vulnerable incumbents that are unable to compete with AI-first entrants or constrained by their massive debt load. This category is problematic for equity and credit investors. Any credit investor who tells you they know who are the winners will be is talking their book. The question remains: how much exposure do you have to the software sector?