Enterprise AI Jumpstart: Start Small, Govern Early, Deliver Value

Artificial intelligence has moved from boardroom buzzword to an operational imperative, yet many organisations remain stalled at the starting line — frozen by cost fears, data readiness questions and governance uncertainty — a gap that demands a practical, measurable path from enthusiasm to execution.

Background: why the narrative matters now​

The conversation about enterprise AI in 2025 is no longer theoretical. Senior leaders routinely describe AI and analytics as central to strategy, and market research projects explosive growth in AI investment and tooling. A Gartner survey found that 79% of corporate strategists view analytics and AI as critical to their success over the next two years, while the same research showed only a minority reporting meaningful operational use of AI-related tools in their functions. This gap — intention versus deployment — is a core driver of what industry voices now call “AI paralysis.”
At the same time, market forecasts amplify the pressure to act. One widely cited MarketsandMarkets projection estimated the global AI market could reach roughly $407 billion by 2027 at a 36.2% compound annual growth rate, figures often cited by vendors and boards when arguing for accelerated AI spend. Those headline numbers are useful for context but do not substitute for a disciplined ROI plan.
Analyst commentary adds another structural caveat: tooling is both improving and proliferating. IDC and other research groups describe an ecosystem that is maturing in capabilities while fragmenting in vendor options and approaches — a dynamic that produces choice fatigue for procurement and engineering teams. In short, leaders face urgency to act and an overload of competing technical paths to choose from.
This feature synthesises those realities, summarises the practical framework urged by Chris Badenhorst of Braintree, and critically examines the strengths and risks of the “start-small, govern-early” approach many partners now offer. It also provides a tactical roadmap for Windows- and Azure-centric IT teams looking to turn AI enthusiasm into measurable outcomes.

---

Overview: three structural causes of AI paralysis​

Organisational interviews and recent industry commentary converge on three recurring causes of paralysis. Each is real, solvable — but only with disciplined choices.

1. Perceived cost and uncertain ROI​

Many leaders assume AI requires massive, upfront investments in GPUs, sprawling data lakes and large teams before any value appears. That “data-lake-or-bust” mental model discourages rapid experimentation and pushes organisations to delay starting until every variable is perfect. The reality is that meaningful pilots can be scoped and measured in weeks or months, not years, if objectives and KPIs are chosen carefully.

2. Data readiness and quality​

Generative models and predictive systems depend on reliable, well-structured data. Organisations often underestimate the effort needed to cleanse, structure and curate the data required for trustworthy AI outputs. Skipping foundational data work increases the risk of poor model performance, biased outputs or costly rework later. Practical pilots are most successful when they constrain scope to the minimum viable dataset required to prove a hypothesis.

3. Security, governance and compliance​

Where data flows, who can access it and how decisions are logged are non-negotiable concerns — especially in regulated industries. Leaders are rightly wary of the legal and reputational risk that comes with ungoverned agentic systems or unmanaged model hosting. Embedding governance, observability and human-in-the-loop checkpoints from day one reduces these risks and protects future scale.
These three pressures combine into a rational risk-avoidance stance: act too quickly and you risk regulatory or financial harm; wait too long and you risk falling behind competitors. The solution lies in clarity — concrete outcomes, tight scope and measurable pilots.

---

From confusion to clarity: a practical enterprise AI framework​

Practical action requires a reproducible operating model, not another technology roadmap. The framework below mirrors the advice from experienced implementers and platform vendors and is intentionally platform-agnostic while mapping readily to Azure- and Microsoft-centric environments.

Start with outcomes, not models​

• Identify 2–3 micro-use cases tied to specific, measurable business KPIs (time saved, error reduction, revenue uplift, NPS change).
• Keep scope small so success can be measured in 60–120 days.
• Ensure the outcome is meaningful enough to justify scale if the pilot succeeds.

Scope for minimal viable data​

• Avoid the “data lake or bust” trap by using the smallest, well-structured dataset needed for the case.
• Use techniques like tenant grounding, prompt filtering and synthetic data to speed test cycles while protecting privacy.
• Design data contracts that enable incremental addition of sources after the pilot proves value.

Bake governance into every pilot​

• Treat identity, access controls, audit trails and rollback procedures as deliverables, not afterthoughts.
• For agentic or decision-making scenarios, enforce least-privilege grants and human-in-the-loop gates on actions.
• Include observability and simple LLMOps controls (latency/error monitoring, data provenance, retraining cadence).

Use managed services to shorten time-to-value​

• Leverage platform-managed capabilities for model hosting, vector stores, identity and telemetry to reduce specialist hiring burdens.
• For Azure-centred shops, this means using Azure AI services, Azure Machine Learning and integrated identity/monitoring tools where appropriate.

Measure, learn and scale incrementally​

• Instrument every pilot with both technical and business metrics.
• Run a 6–12 month measurement window before scaling broadly.
• Convert successful pilots into templated patterns (identity, data access, monitoring) to accelerate replication across domains.

---

The Braintree prescription: Azure AI Jumpstart and partner-led readiness​

Braintree positions its Azure AI Jumpstart as a structured readiness programme that accelerates the first steps: readiness assessment, micro-pilot selection, proof-of-value and governance playbooks. For organisations already invested in Azure and Microsoft 365, that provider-aligned approach reduces integration friction and provides a clear route to tie pilots directly to outcomes. Braintree’s published materials, Azure Marketplace listings and marketing show a portfolio focused on data readiness, Copilot assessments and rapid pilot delivery.
What a typical Azure-aligned Jumpstart delivers in practice:

• Rapid baseline assessment of data maturity and identity posture.
• A narrowly scoped pilot, measurable in weeks.
• Governance and FinOps templates that can be operationalised.
• Skills-transfer plans and optional managed service handover.

These deliverables match the practical, outcome-first pattern recommended by corporate IT teams and major platform vendors. However, feasibility depends on the partner’s depth of Azure IP, vertical accelerators and post-pilot handoff quality. Buyers should demand evidence of prior success and contractual portability protections.

---

Strengths of the start-small, govern-early model​

• Lower upfront cost: Pilot-first approaches avoid heavy initial infrastructure spend and allow TCO to be measured against demonstrable benefits.
• Faster buy-in: Measurable wins create internal champions and reduce resistance to scaling.
• Governance-as-default: Building controls into the pilot decreases governance debt and mitigates regulatory exposure.
• Leverage platform investments: Organisations entrenched in Azure can reuse identity, monitoring and compliance patterns for faster time-to-value.

---

Where the model can fall short: vendor lock-in, incomplete handoffs, and measurement traps​

A sensible pilot programme reduces risk — but it does not eliminate it. The common blind spots include:

• Vendor lock-in and portability risk. If pilots are built on proprietary hooks without portability clauses, scaling can become costly and strategically constraining. Procurement must demand rollback and portability clauses as a condition of engagement.
• Insufficient handoff to internal teams. Some partners deliver pilots but fail to transfer operational knowledge. Ensure the engagement includes retraining plans, documentation and shadowing to institutionalise capabilities.
• Measurement theater. Anecdotal productivity claims are compelling but need rigorous before/after measurements and control groups to be credible. Treat single-case ROI numbers as illustrative until verified.
• Neglecting total cost of ownership (TCO). Fast pilot success can mask runaway inference costs or retraining spend at scale. Enforce FinOps discipline from the start and require transparent pricing scenarios for training, storage and inference.

These limitations are manageable if teams apply procurement discipline, demand demonstrable metrics and treat the pilot as the start of an engineering project rather than a marketing exercise.

---

Microsoft Copilot: why it matters — and why it’s not the whole story​

Microsoft’s Copilot product family has been effective at lowering the psychological barrier to AI by embedding generative capabilities into apps people use daily. Copilot for Microsoft 365 and Copilot in Teams, Outlook and PowerPoint make AI tangible for knowledge workers and accelerate early adoption by normalising natural-language interactions inside familiar workflows. Microsoft’s product messaging and in-product features have explicitly prioritised productivity scenarios, which helps build momentum for larger programmes.
But Copilot is just one layer of enterprise AI. It can automate drafting, summarisation and routine tasks, yet it does not replace the need for strong data engineering, observability, model lifecycle management and governance controls for business-critical or high-risk workloads. Organizations should treat Copilot adoption as an early engagement play that can lower resistance, while building the engineering and compliance backbone needed for mission-critical use cases.

---

A disciplined procurement checklist for IT leaders​

Procurement and legal teams must be active participants in early AI pilots. The following checklist converts vendor conversations into accountable contracts:

• Require transparent pricing scenarios covering training, inference and storage across realistic utilization patterns.
• Demand evidence of data residency, encryption and breach-notification policies.
• Insist on rollback, portability and observability clauses to avoid being stranded on a single model/provider.
• Ask for SLAs around latency, error rates and retraining windows.
• Require demonstrable customer references and success metrics from similar pilots.
• Include FinOps guardrails and monthly production cost reporting.

This checklist protects finance, security and product teams while keeping the focus on repeatable outcomes.

---

Tactical playbook: nine operating steps to move from paralysis to progress​

• Define one narrow business objective that AI will serve (not generic “deploy Copilot”).
• Run a 60–120 day pilot with clear KPIs and a defined measurement approach.
• Use a customer-zero approach to pilot governance and security before external rollout.
• Establish a lightweight centre of excellence to provide templates and reuseable integrations.
• Build an adoption program with champions and daily practice sessions.
• Instrument both technical telemetry and business KPIs for every pilot.
• Reclaim unused seats and licenses proactively to avoid shelfware.
• Maintain human-in-the-loop checkpoints for high-risk decisions.
• Convert pilot playbooks into scalable programs and feed lessons back into CoE standards.

This sequence balances speed with prudence, enabling measurable value capture while preserving governance and financial discipline.

---

Evidence and verification: how claims hold up under scrutiny​

Several claims in public discussions deserve close verification before they shape large-scale decisions:

• The Gartner statistic that 79% of corporate strategists view AI as critical and that only 20% reported using AI-related tools in their function is accurate to the Gartner press release. However, the shorthand “only 20% use AI every day” conflates two different measures — the survey reported 20% of strategists reported using AI-related tools for their function, not explicitly daily usage — so leaders should treat the “daily use” framing with caution and read the underlying methodology.
• The MarketsandMarkets $407 billion by 2027 estimate and 36.2% CAGR is a published market forecast originally released in 2022 and widely circulated; it is a valid market projection but should be used as directional context rather than a precise budgeting target. Market forecasts vary widely by scope, methodology and date of publication.
• IDC’s commentary on ecosystem dynamics supports the core idea that tooling is both maturing and fragmenting; while the exact phrasing “simultaneously maturing and fragmenting” is a useful summary, it is effectively a synthesis of multiple analyst observations rather than a single IDC soundbite — treat it as a sector characterization rather than a data point.

Where claims are anecdotal — for example, specific client ROI percentages or singular “30% workload reduction” stories — they should be treated as illustrative unless the vendor can supply the underlying before/after metrics and methodology. Demand transparency on measurement.

---

For Windows-centric IT teams: practical, immediate steps​

• Pilot Microsoft 365 Copilot features that automate repetitive tasks on the user desktop, and measure time-saved metrics using admin telemetry to build an evidence base.
• Standardise a secure pattern for Copilot: tenant grounding, prompt filtering and conditional access policies. Treat Copilot-generated artifacts as you would any business record (retention, eDiscovery).
• Integrate Copilot telemetry into your monitoring and SIEM stacks to track adoption and output quality, and invest in developer enablement so Windows, Azure and MLOps teams can collaborate on LLMOps patterns.

---

Final assessment: where clarity wins​

The central argument is straightforward: scarce enterprise resources are not best spent chasing the next model headline; they are better invested in building clarity — concrete outcomes, measurement discipline, governance and repeatable operational patterns. The AI market will continue to expand and fragment; vendors will iterate rapidly. Organisations that convert enthusiasm into tightly scoped experiments, insist on evidence-based measurements and contractual safeguards, and build the operational muscles (data, identity, observability, FinOps and governance) will win. Microsoft Copilot and other in-app copilots reduce the psychological barrier to entry, but they do not remove the need for engineering rigour and compliance frameworks when AI moves into business-critical domains.

---

Closing checklist for leaders (quick reference)​

• Map 2–3 high-value micro-use cases with explicit KPIs.
• Run a 30–90 day rapid data health check for those cases.
• Draft a governance playbook (identity, audit, rollback).
• Run a small POC on representative data before procurement.
• Define FinOps metrics and cost transparency before scaling.
• Enforce a 6–12 month measurement window and formal decision gate for scale.

Clarity is, in many organisations, the scarce commodity. It trumps model count, marketing noise and vendor FOMO. Start small, measure precisely, govern strictly and scale deliberately — and AI will shift from a source of anxiety into a predictable engine of value.

---

Source: IOL Changing the AI narrative from confusion to clarity

 

[ChatGPT]

SaaS vendors are quietly — and sometimes not so quietly — raising the effective cost of cloud software well ahead of inflation, and organisations that assume price increases are inevitable are already losing the fight for predictable, sustainable software spend. Analysts at Gartner told attendees at recent Symposium events that list-price uplifts and a wave of pricing model changes — from token/credit schemes and repackaging to AI-specific surcharges — have combined to push SaaS bills into double-digit territory for many customers. The good news: procurement, IT and finance teams armed with the right data and a two‑year playbook can blunt much of that pain, force transparency, and in many cases avoid paying the full proposed uplift.

Background / Overview​

SaaS pricing used to be simple: a per‑seat fee and maybe a storage or over‑usage line item. That era is over. Vendors are layering new metrics — tokens, credits, agent charges, AI compute units — onto subscription contracts and, in parallel, repackaging bundles so that previously included features become paid add‑ons. At industry events this year, Gartner analysts flagged a broad range of practices vendors are using to increase revenue, from modest list increases to wholesale metric changes that multiply customer bills overnight. Meanwhile, vendors argue these changes fund necessary investments — particularly around generative AI — and also reflect currency effects and rising infrastructure costs. Buyers see a different reality: contracts that are harder to compare, more hidden costs, and renewal cycles that are becoming battlegrounds.
This is not theoretical. Major enterprise platforms have adjusted prices and licensing metrics in ways that materially affect customers, and legal battles and regulatory scrutiny have followed some of the most aggressive moves. Such examples have prompted procurement teams to rethink negotiation timing, contract language and technical controls to prevent surprise invoices.

---

What’s actually changing in SaaS pricing?​

New pricing levers vendors are using​

• Token / credit systems: Vendors sell packages of credits that are consumed by actions (API calls, model tokens, premium feature usage). Vendors often reserve the right to change credit-to‑unit multipliers, effectively doubling or tripling unit prices without an explicit list price change.
• AI surcharges and bundled AI features: Generative AI features are commonly positioned as premium value-adds. Vendors either bolt AI into higher tiers or create separate AI SKUs with substantially higher per‑unit costs.
• Repackaging / unbundling: Features that were part of standard tiers get moved into higher tiers or sold as paid add‑ons. This bundling shuffle forces many customers to upgrade.
• Metric changes (API, per‑GB, per‑agent): Vendors shift from seat-based licensing to consumption metrics such as CPU cores, API calls, tokens, or model inference units — metrics that are harder to forecast.
• Mandatory upgrades and feature retirements: Vendors sometimes retire an older, cheaper product in favour of an AI‑enhanced replacement that costs significantly more per user.
• Minimum commitments and term locking: Pressure to accept longer terms or larger minimum spends in exchange for nominal discounts — which locks customers in before further price resets.

Why these changes hit so hard​

Many organisations buy SaaS in siloes: product teams pick specialist tools, helpdesk purchases are handled by operations, and business units sign up for third‑party apps without central oversight. When renewals arrive, procurement often discovers they’re negotiating on a whack of disconnected accounts, with no single view of usage, overlapping licenses, or common leverage points. That creates perfect conditions for vendors to raise prices piecemeal and extract additional revenue.

---

Why vendors are accelerating price changes​

Funding AI and infrastructure​

Vendors point to rising infrastructure and R&D costs, especially for generative AI: model training, inference, and inference latency SLAs require expensive GPU infrastructure or specialised cloud services. Recouping that spend through new pricing lines is an obvious route.

Margin pressure and market signalling​

The enterprise software market has evolved: private and public vendors feel pressure to show growth and expand margins. Price changes are a straightforward lever. The Broadcom acquisition of VMware (and downstream pricing overhauls) is a bellwether: firms watch one vendor successfully monetise legacy assets and often question whether they can follow suit.

Strategic product positioning​

By moving advanced features behind higher tiers, vendors can force customers to consolidate spend or migrate to new integrated stacks — a strategy to increase per‑account revenue and stickiness.

---

The “VMware effect” and regulatory pushback​

When a major vendor resets pricing and packaging aggressively, the ripple effects are immediate. Since Broadcom’s acquisition of VMware, multiple customers and public bodies have reported steep renewal uplifts and product repackaging that forced re‑evaluation of long‑standing deployments. In some cases customers have claimed proposed uplifts hundreds of percent higher than prior spend; others have taken legal action or sought regulatory remedies to secure migration support or to block sudden terminations.
This high‑profile set of disputes has two important consequences for buyers:

• It demonstrates the worst‑case for vendor leverage: customers with entangled, mission‑critical deployments face very high migration costs and operational risk.
• It has focused regulator and industry attention on licensing conduct, leading some large customers and trade groups to seek remedies or scrutiny.

That said, not every vendor will — or can — repeat the most extreme paths. The debate has merely made it clearer that buyers must assume vendors have significant leverage and prepare accordingly.

---

The hard numbers (what we can verify)​

• Industry studies and procurement datasets show SaaS pricing has risen materially across many categories; independent reports have documented average increases in the low‑to‑teens percentage range year‑over‑year for many vendors, and a mix of much larger uplifts on specific SKUs or for customers who did not renegotiate early.
• Analyst forecasts for overall IT spending show budgets are growing, but not uniformly or fast enough to absorb all vendor price increases. That means procurement must prioritize efficiency and rightsizing to prevent runaway consumption from eroding discretionary IT budgets.
• Public legal filings and press accounts validate extreme anecdotes — including multi‑hundred‑percent suggested uplifts for specific enterprise customers — even if those represent the high end of the distribution.

Where claims differ, or a specific percent is quoted for a broad vendor cohort (for example, “SaaS vendors hiked prices 9–25% this year”), those figures reflect analyst briefings and sampling rather than a single global index. Buyers should treat headline percentages as directional and focus on the concrete terms of their own contracts.
Note: some specific percentage claims circulating in commentary are hard to verify across an entire market; they are best used to illustrate scale rather than exact averages.

---

How procurement and IT can push back — a two‑year playbook​

The central insight analysts stress is simple: start early and plan broadly. Waiting until a renewal notice appears leaves you at a severe disadvantage.

Year −2 (24 months before renewal): Discover and consolidate​

• Inventory everything — Create a single source of truth for all SaaS subscriptions, contracts, SKUs, renewal dates, usage metrics, and owners.
• Map functional overlap — Identify where the same vendor or feature is bought by separate teams (CRM, chat, ESB, analytics) and tag opportunities for consolidation.
• Classify value — Label licenses by criticality and value delivered: business‑critical, useful, low‑value/replaceable, or unused.
• Begin migration planning — If a product is vulnerable to a large proposed uplift, map migration options (in‑house, alternate vendors, open source) and estimate migration cost and timeline.

Year −1 (12 months before renewal): Build leverage​

• Engage vendors early — Tell vendors you are evaluating options, and ask for transparent pricing models, calculators, and historical consumption data.
• Demand pricing mechanics in contract — Push for fixed multipliers or caps, clear definitions of credits/tokens, and advance notice for any metric changes.
• Consolidate spend where sensible — Combine related contracts across business units to increase bargaining power.
• Prepare RFPs and pilots — Run competitive bids for high‑spend areas. Create proof‑of‑concept migrations for the riskiest and most expensive contracts.
• Legal: negotiate exit and continuity clauses — Require data export rights, export formats, timeline guarantees, and establish data escrow if vendor arbitrarily terminates service.

6 months before renewal: Negotiate hard​

• Use migration costs as a bargaining chip — Present credible migration estimates; vendors often prefer discounts to the reputational and churn costs of a migration.
• Ask for cross‑product discounts — If you buy multiple products from the same vendor, package them into a single negotiation for better pricing.
• Demand forecasting tools — Get vendor commitment to provide cost calculators and monitoring for consumption‑based metrics before renewal.
• Design SLA/penalties for sudden metric changes — Seek contractual remedies if pricing definitions are altered mid‑term.

Renewal and post‑renewal: Operationalise protection​

• Monitor consumption continuously — Use FinOps and SAM tools to detect spikes, token burn rates, and unexpected multipliers.
• Rightsize and reassign seats — Regularly audit seat usage and migrate heavy users to the tiers or tools that make economic sense.
• Run “what‑if” stress tests — Model cost scenarios with varying multipliers and AI adoption curves; budget for the actual operational cost of enabling AI agents.
• Exercise exit rights when necessary — If continuance would be untenable, execute the migration plan proactively rather than being forced into poor terms.

---

Practical negotiation levers that work​

• Consolidation discounts: Combine spend across a vendor’s portfolio for scale discounts.
• Multi‑year commitments with explicit escape clauses: Consider offering term length in exchange for price caps and migration assistance.
• Customer reference commitments: Offer to act as a reference or to participate in case studies in exchange for better renewals.
• Usage gates and knockouts: Contractually fix how credits convert to billable units, and include a clause requiring 90+ days’ notice before the vendor can change multipliers.
• Audit and transparency rights: Right to inspect vendor billing logic and reconciliation reports.
• Data portability and escrow: Define format, frequency and costs for data export and escrow provisions to avoid vendor hold‑ups during migration.

---

Technical and organisational controls: how to limit surprises​

• Implement a centralised SAM (Software Asset Management) and FinOps practice that combines usage telemetry with procurement records.
• Instrument APIs and telemetry to count actual consumption units (API calls, tokens, CPU hours) and set programmatic alarms when thresholds are approached.
• Create cost‑aware developer and product teams: expose cost per action in developer consoles so building features is not blind to operational cost.
• Invest in cost modelling tools and insist vendors provide their pricing calculators before signing.
• Use feature toggles and staged rollouts for AI features so adoption is measured and budgeted, not accidental.

---

Legal and geopolitical risks: don’t overlook them​

Vendors retain rights in many standard contracts to suspend service under sanctions or force majeure. That can leave you vulnerable if a vendor quietly runs data through regions or providers with geopolitical exposure. Practical steps:

• Require contractual disclosure: Where are your workloads hosted? What regions are used for backups? Insist on notification of geographic changes for storage and processing.
• Negotiate continuity / migration support: If your vendor terminates access for legal reasons, require that they provide assistance (data export, APIs, transfer windows) that allows you to continue operations.
• Test recovery: Conduct dry runs that move key workloads between regions or simulate export/import operations so you know the true time and cost.

---

Weighing the tradeoffs: when to pay, when to walk​

There is no single answer. Each negotiation is a balance between three factors:

• Migration cost and risk — How long and expensive is it to leave?
• Strategic value — How critical is the vendor’s offering to core business functions?
• Vendor behaviour and market alternatives — Are there credible substitutes or enough time to build alternatives?

When migration risk is high and the vendor demonstrates reasonable commercial conduct, paying a negotiated uplift might be pragmatic. When a vendor demands punitive increases or moves terms that unduly concentrate risk, a firm migration posture and readiness to exit can unlock better pricing or contractual concessions.

---

Strengths and justifications vendors raise — and where buyers should be skeptical​

• Valid vendor points: AI and infrastructure investments are real; maintaining global, low-latency services with predictable SLAs costs money. Vendors that transparently show how new fees fund improved functionality or efficiency present a defensible case.
• Where to be skeptical: Sudden metric changes without clear definitions and calculators, or credit systems that allow unilateral multipliers, are commercial red flags. Similarly, when a vendor shifts core features into a new paid tier while discontinuing legacy products, buyers should demand migration assistance or time‑bound grandfathering.

---

A short, practical checklist for the next 90 days​

• Centralise your SaaS inventory and tag all renewal dates.
• Identify top 10 cost drivers and build simple migration cost models for each.
• Start consolidation conversations with vendors that hold multiple product lines in your estate.
• Require pricing calculators and transparent billing logic before signing any extensions.
• Update contract templates to include data export obligations, multiplier caps, and notice periods for pricing metric changes.
• Schedule a simulated migration of a non‑critical workload to validate estimated timelines and costs.

---

Conclusion​

SaaS vendors are entering a new phase of pricing complexity. The combination of AI‑driven product advances and aggressive commercial packaging means subscription bills can rise faster than budgets and far quicker than many organisations expect. But the situation is not hopeless.
The most effective defence is a proactive one: inventory your estate, plan renewals at least two years out, consolidate where you can, demand clarity on the mechanics of consumption billing, and make migration readiness a real option rather than a theoretical threat. When procurement and IT act early and together — armed with data, a credible migration plan, and tight contractual protections — organisations can wrest back control over SaaS costs. Those that don’t will likely face an accelerating cost curve that erodes their IT budgets and reduces strategic optionality.

---

Source: theregister.com SaaS vendors are hiking costs, fast, but you can push back