Enterprise Software Volatility Might be an Investor's Friend

In a few years we might look back and discover some investors were pretty significant winners; others losers as investors in enterprise software stocks, given the current volatility.

If software valuation multiples permanently contract lower, bears will have been proven correct. But if AI disruption proves less damaging, then enterprise software leaders are on sale.

According to Goldman Sachs analyst Matthew Martino, investors need to consider six different aspects of how artificial intelligence might affect an enterprise software supplier’s business:

• Orchestration risk: The possibility that horizontal AI agent layers can bypass the platform and become the primary generator of value

• Monetization model: Whether a business model is tied to users, which makes it more vulnerable, or assets and data, making it more durable

• System-of-record ownership: If the platform governs approvals, compliance, and execution, it is harder to displace

• Data and integration moat: Whether workflows depend on proprietary signals, structured data, and operational records that live inside the platform and must be accessed through it

• AI execution: Whether the company is delivering real, embedded capabilities rather than conceptual roadmaps

• Budget alignment: Determines whether AI adoption increases or decreases the strategic priority of the category. 

For investors, the key takeaways might be:

• Pure workflow or UI-heavy tools are more exposed; systems handling intricate enterprise logic less so

• Subscription models linked to business scale rather than headcount tend to hold up better as AI automates tasks

• Core ERP, financial systems, CRM with compliance hooks, or regulatory platforms are harder to displace

• AI agents amplify the value of clean, governed, real-time data rather than replacing it

• Execution separates fast followers who reinforce their moat from those getting disrupted. Incumbents with domain data have an advantage here

• Look for categories where AI creates tailwinds, like data platforms, cybersecurity, vertical SaaS with complex compliance, or physical-to-digital workflows.

Company	Orchestration Risk	Monetization Model	System-of-Record Ownership	Data & Integration Moat	AI Execution	Budget Alignment
Salesforce (CRM)	Moderate (−) Agentforce agents could be orchestrated externally, but platform workflows limit full bypass	Strong (+) Shifted to agentic ELAs + consumption (per conversation/action); Agentforce ARR ~$800M+ and growing fast	Strong (+) Core CRM is authoritative for customer data, approvals, compliance	Strong (+) Deep customer interaction history, proprietary signals, Data Cloud integrations	Strong (+) Shipping real agents (Agentforce) with measurable ROI; Einstein embedded	Strong (+) AI boosts CRM priority & spend (more data, governance, agent infra)
ServiceNow (NOW)	Moderate-Strong (+) Complex, multi-system IT/service workflows hard for horizontal agents to fully orchestrate	Strong (+) Now Assist “Pro Plus” premium tiers + consumption upside; ACV >$600M–$1B run-rate	Strong (+) System-of-record for IT/service management, approvals, audit trails	Strong (+) Proprietary operational workflows + deep integrations across enterprise tools	Strong (+) Now Assist agents delivering real value; infrastructure fully cloud-native for AI scale	Strong (+) AI adoption increases strategic priority of service/ops platforms
Oracle (ORCL)	Strong (+) Deep ERP/supply-chain/finance processes resist simple agent bypass	Moderate-Strong (+) Mix of license + cloud consumption; moving toward outcome-based	Strong (+) Authoritative ERP/finance/supply chain system-of-record with compliance hooks	Strong (+) Massive proprietary enterprise data across finance, supply chain, HR	Strong (+) Using AI to build full automated processes (not just copilots); ahead on domain execution per some peers	Strong (+) AI tailwinds for database/cloud infra spend
Workday (WDAY)	Moderate (−) HR/finance copilots can partially bypass UI; more replicable workflows	Moderate (−) Still largely seat/user-tied; slower shift to consumption	Strong (+) Core HR/finance system-of-record for employee data & compliance	Moderate (−) Data is relatively standardized vs. highly proprietary; easier for AI to replicate externally	Moderate (+) Has copilots but viewed as lagging some peers on full agentic execution	Moderate (+) AI helpful but could shift some budget away from traditional HCM suites
SAP (SAP)	Strong (+) ERP complexity & regulatory workflows make full orchestration bypass difficult	Moderate-Strong (+) Cloud shift + some consumption elements; historical pricing power	Strong (+) Dominant global ERP system-of-record for finance, supply chain, compliance	Strong (+) Deep structured operational data + decades of integrations	Moderate-Strong (+) Embedding AI into core ERP; fast-follower re-architecture	Strong (+) AI increases need for governed ERP data & execution layers
Adobe (ADBE)	Moderate (−) Generative tools could let agents bypass some creative workflows/UI	Moderate-Strong (+) “Generative Credits” + consumption model for Firefly; shifts value capture to usage	Moderate (+) Creative asset system-of-record but less regulatory/compliance weight	Moderate-Strong (+) Proprietary creative workflows + Firefly-trained data moat	Strong (+) Firefly & generative AI delivering real embedded capabilities; strong execution in content	Moderate-Strong (+) AI boosts content creation budget but compresses some per-seat value
Snowflake (SNOW)	Moderate (−) Data platform can be bypassed as pure backend; agents query elsewhere	Strong (+) Consumption-based (compute/data usage); Cortex AI adds monetization vectors	Moderate (+) Strong data platform but not always the business process system-of-record	Strong (+) Data cloud moat + governance; AI amplifies need for clean, governed data	Strong (+) Cortex AI services shipping real capabilities on customer data	Strong (+) AI dramatically increases data volume/query spend (force multiplier)
MongoDB (MDB)	Moderate (−) Flexible document DB can be commoditized by agents using open formats	Strong (+) Consumption-based usage; vector search/AI integrations drive upside	Moderate (+) Application data store but rarely the final system-of-record	Moderate-Strong (+) Developer-friendly data + vector capabilities; less proprietary than ERP data	Moderate-Strong (+) Vector search & AI app tools executing well; benefits from AI dev boom	Strong (+) AI apps drive more database consumption and developer spend

And even if all that is correct, the palpable investor fears about how to value enterprise software firms do not seem to be abating. But some analysts think a “new normal” valuation level is close to stabilizing.

Software Segment	New "Floor" P/E	Historical Context
Mega-Cap SaaS (Microsoft, SAP)	28x – 32x	Historic: 35x+
High-Growth / "Rule of 40" (ServiceNow, CrowdStrike)	45x – 55x	Historic: 80x – 100x+
Mature / Cyclical Enterprise (Salesforce, Oracle)	18x – 24x	Historic: 25x – 30x
Infrastructure / Dev Ops (Datadog, Snowflake)	50x – 60x	Historic: 100x+
Mid-Market / "Broken" SaaS	12x – 16x	Historic: 25x

In 2021, software was valued on Enterprise Value/Revenue (often 15x–20x). 

Analysts now believe the stable floor is EV/Free Cash Flow or Forward P/E. 

For a standard healthy software company, a 5x–6x revenue multiple is now considered "stable," whereas 10x or more is reserved only for elite AI-winners.

Analysts at firms including Goldman Sachs and Morgan Stanley believe "legacy" enterprise software could fall further (P/E of 12x–15x) if they cannot prove AI utility.

Based on those estimates, there is considerable danger of further downside.  

Company	Est. Forward P/E (FY 2026)	Trend vs. 5-Year Average	Valuation Driver
Microsoft	22.4x – 23.2x	~30% Compression	Balanced by Azure AI growth and office productivity dominance.
Adobe	23.5x – 25.0x	~20% Compression	Premium maintained due to Firefly AI integration in Creative Cloud.
Salesforce	20.8x – 22.0x	Significant Reset	Shift toward margin expansion and buybacks over aggressive M&A.
Oracle	18.8x – 19.5x	Relatively Stable	Lifted by OCI (Oracle Cloud Infrastructure) demand for AI training.
SAP	21.0x – 22.5x	~15% Compression	Resilient due to "sticky" ERP migration to S/4HANA Cloud.
Workday	43.1x – 43.5x	>50% Compression	Transitioning from high-growth SaaS to a mature HCM platform model.

Analysts at J.P. Morgan and Bessemer note that legacy firms are now strictly valued on their ability to maintain a combined growth and profit margin of 40 percent (“the rule of 40”). Firms falling below this are seeing P/E multiples dip into the mid-teens.

Also, private consensus is that software multiples will not return to 2021 levels unless the 10-Year Treasury falls below three percent. As long as rates remain "higher for longer," analysts are modeling a permanent 20-percent haircut on terminal valuation multiples compared to the last decade.

source: MacroMicro 

Why Video Streaming Can be Much More Profitable Than Music Streaming (for Distributors)

Even if there are similarities for distributors in the streaming video and streaming music businesses, for most entities, if there was a choice, you’d probably choose to be in the video business, not music. 

Music streaming is good for copyright holders but pretty difficult for distributors, while video streaming is better for distributors at scale, and less favorable for copyright holders. 

Both types of streaming share core digital economics: high upfront fixed costs for content creation or licensing, followed by near-zero marginal costs for additional distribution.

But they diverge sharply in how revenue flows to copyright holders (artists/labels/studios) and distributors. 

The marginal cost of streaming additional songs is linear: play a stream, pay a fee. Video streaming is different: content is typically licensed for a flat, fixed fee covering unlimited streams.

So video streamers can reach higher margins as additional subscribers are added: the marginal cost comes mostly in the form of marketing or acquisition cost, not content rights payments. 

Music streamers, on the other hand, pay 70 percent of revenue to copyright holders, at the margin. Volume helps, but only so much. 

source: Joel Goveia 

Platforms such as Spotify pay out 70 percent of revenue to rights holders (roughly 55 percent to 60 percent to labels/masters and 10 percent to 15 percent to publishers). So distributor costs are variable with scale. 

More streams mean higher payouts.

Video streamers pay flat fees for licensing content, so digital scale economies work. 

For a video streamer, there is no per-view royalty. Netflix’s costs, for example,  are largely fixed upfront (production or licensing deals), so additional views do not increase payments to rights holders.

Video streaming licensing also means differentiation is possible. Virtually all music streamers have access to the same content.

Video licensing is restricted: content can be supplied uniquely on a single platform. Also, some video streamers (such as Netflix) can create original content and own it. 

Aspect	Similarities	Music Streaming – Copyright Holders	Music Streaming – Distributors (e.g., Spotify)	Video Streaming – Copyright Holders	Video Streaming – Distributors (e.g., Netflix)
Primary Revenue to Holders	Subscription-driven platform access	Revenue share (~55-60% masters, 10-15% publishing from platform revenue)	Pays ~70% of revenue to holders (variable)	Fixed lump-sum licensing fees or IP ownership	Fixed licensing or owns originals (no per-view royalties)
Marginal Cost Structure	Near-zero for delivery (bandwidth/storage)	Proportional to streams (recurring royalties)	Variable costs rise with usage/revenue	Upfront fixed fee (unlimited plays)	Fixed after acquisition; bandwidth only
Content Ownership	Holders license access rather than sell copies	Retain rights; perpetual licensing	No ownership; pure licensing	Retain/sell licensing rights; strong IP control	Increasingly owns originals for future value
Consumption Incentive	Scales with catalog size and user engagement	`` ++	Wants high engagement but pays more for it	Favors binge/complete viewing	Benefits from high consumption of sunk-cost content
Risk & Profit Model	High fixed costs; economies of scale at large user base	Stable but diluted per-stream payouts	Thin margins; freemium helps acquisition	Front-loaded, predictable fees	Higher margins possible; originals drive differentiation
Business Model Type	Platform economy with low marginal costs	Per-stream / pro-rata royalties	Revenue-share licensing	Flat-fee licensing or ownership	Fixed-cost licensing + owned content

The bottom line is that if an entity has a choice, it will want to be in the video streaming business rather than music streaming. 

The caveat is that some entities use music and/or video streaming as a “loss leader” to add value for some other product feature that actually drives the direct profits and profit margins. Amazon Prime video and music provide a good example.

Amazon AI Capex is a Rorschach Test

Amazon capital spending plans are a bit of a Rorschach test test these days: what the picture means tells us more about the viewer’s perceptions than the objective facts about the investments. 

figures in millions 

Investors and analysts fear the impact on free cash flow, of course. But Amazon operating margins still look decent, at the moment. 

Amazon CEO Andy Jassy's shareholder letter might put the firm’s thinking into perspective. Jassy says that although “reasonable people can disagree,” “when you identify disproportionate inflections, bet big.”

And Amazon believes AI is that sort of thing. 

“Three years after AWS launched commercially, it had a $58 million revenue run rate,” Jassy says. “Three years into this AI wave, AWS’s AI revenue run rate is over $15 billion in Q1 2026, nearly 260 times larger than AWS at that same point.”

And everyone agrees there now is excess demand for high-performance computing capability. “We still have capacity constraints that yield unserved demand,” says Jassy. As an example, he notes that ”two large AWS customers have already asked if they could buy ‘all’ of our Graviton instance capacity in 2026 (custom CPU chip).”

And it matters how big, and how fast, the new business grows. “The way AWS’s cash cycle works is that the faster AWS grows, the more short-term capex we’ll spend.” Everyone understands that. 

But that’s where the Rorschach analogy kicks in. “The free cash flow and return on invested capital  for these investments are cumulatively quite attractive a couple years after being in service,” Jassy says.

“However, in times of very high growth (like now), where the capex growth meaningfully outpaces the revenue growth, the early-years FCF is challenged until these initial tranches of capacity are being monetized and revenue growth out-paces capex growth,” Jassy says. 

What investors and analysts “see” in the ink blots is conditioned by their expectations about the size of the opportunity; the profit margins; the payback period and the likelihood that a “winner takes most” market structure will develop. 

“Every customer experience will be reinvented by AI, and there will be a slew of new experiences only possible because of AI,” Jassy notes. 

Agree or not, his views on a few key questions are emphatic”

• Is the technology over-hyped

• Are we  in “a bubble”

• Will the margins and ROIC will be appealing. 

“My strong conviction, at least for Amazon, is that the answers are no, no, and yes,” says Jassy. 

Strong opinions are held for each of those questions, and investor bets will follow. Big gains or big losses are possible. But investor decisions on where to place bets are a Rorschach test.