The 10-Year Treasury Is Above 5%. Time to Reset the Hurdle Rate on Your AI Projects | MMARW
MMARW / INTELLIGENCE / FINANCE
The 10-Year Treasury Is Above 5%. Time to Reset the Hurdle Rate on Your AI Projects
The 10-year Treasury touched 5.34% intraday on October 1 and is still above 5%. AI projects approved on 2025 math need a second look, especially the ones that pay back late.AI-assisted publicationAI contributed to the research, drafting, or imagery. MMARW retains editorial responsibility for the published page.
11 min readPublished By MMARW
AI
The 10-year Treasury above 5%: time to reset AI hurdle rates. · MMARW
As long-term US Treasury yields breach multi-decade highs, the capital allocation models used to greenlight AI investments in 2025 are becoming obsolete; CFOs must now re-evaluate the cost of capital to protect margins.
Thesis
The rapid ascent of long-term US Treasury yields has fundamentally altered the cost of capital for corporate projects. Throughout 2025, many Artificial Intelligence (AI) initiatives—ranging from foundational data infrastructure to generative workflow automation—were approved based on hurdle rates calibrated to a much lower risk-free rate environment. With the 10-year Treasury yield now consistently testing levels above 5%, the Net Present Value (NPV) of long-dated AI investments is being eroded by the higher discount rate required to compensate for the increased opportunity cost of capital.
Management cannot afford a blanket approach to this shift. A uniform increase in the hurdle rate across all AI projects will result in "false negatives," where strategically vital, long-term transformations are cancelled because they cannot meet an arbitrary new threshold. Conversely, failing to adjust hurdle rates will lead to "value destruction," where capital is sunk into multi-year projects that no longer generate returns relative to the risk-free benchmark.
The recommendation is a disciplined reset: re-calculate the Weighted Average Cost of Capital (WACC) using current yields, re-rank the existing AI project pipeline based on these new rates, and apply differentiated hurdle rates based on project duration and risk profile. Specifically, long-payback infrastructure plays must face higher scrutiny, while short-payback automation may remain viable. If the 10-year yield stabilizes or begins a significant downward trend, the decision criteria should be reverted to prevent under-investment in critical technological capabilities.
What actually happened
The landscape for long-term capital costs shifted dramatically in late 2026. On 1 October 2026, the 10-year Treasury yield reached an intraday high of 5.342%, surpassing its 2007 peak and marking its highest level since early 2002 Reuters. Although it retreated slightly from that intra-day peak, the 3pm close on 1 October sat at 5.233% .
This rise reflects a volatile period for fixed income. The 10-year yield saw a 52-week high close of 5.292% on 30 September 2026 and has trended upward significantly, rising approximately 1.08 percentage points year-to-date Morningstar / Dow Jones Data Talk. The daily par yield curve data from the U.S. Treasury confirms this volatility: the 10-year stood at 5.29% on 30 September, 5.24% on 1 October, and 5.31% on 5 October 2026 U.S. Treasury daily par yield curve rates.
Critically, the current yield curve exhibits a structure where long-term rates are significantly higher than policy rates. As of 8 October 2026, the 10-year yield was 5.22%, while the 2-year was 4.75%, the 5-year was 4.99%, and the 30-year was 5.60% U.S. Treasury daily par yield curve rates. This indicates that the pressure on long-term project discount rates is being driven by term premiums and inflation expectations rather than just immediate central bank policy.
Why it matters for the reader's decision
For a CFO or COO, the link between Treasury yields and AI project approval is not theoretical; it is a mathematical certainty governed by the mechanics of the Weighted Average Cost of Capital (WACC).
When evaluating a capital expenditure (CapEx)-heavy project like an enterprise-wide AI data platform, firms use WACC to determine the minimum acceptable return, known as the hurdle rate. The WACC is composed of two primary segments: the cost of equity and the cost of debt.
The Risk-Free Rate (Rf): This is the foundation of both components. The cost of equity is typically calculated using the Capital Asset Pricing Model (CAPM): Cost of Equity = Rf + β × (Equity Risk Premium). As the 10-year Treasury yield—the standard proxy for the risk-free rate—rises, the required return for equity holders rises proportionally.
Cost of Debt: While corporations may have existing fixed-rate debt, the marginal cost of new debt (funding new AI hardware or cloud expansions) tracks current market yields.
The Beta (β) and Risk Premium: AI projects often carry higher operational risk than standard software updates. This risk is captured by the beta coefficient. Even if beta remains constant, a higher Rf increases the absolute dollar amount of the risk premium required to justify the investment.
The timing of these cash flows is the "killer" for AI projects. Because AI transformations are frequently "front-loaded" (massive initial investment in compute and data cleaning) with "back-loaded" benefits (efficiency gains realized 3–5 years later), they are hyper-sensitive to the discount rate. A higher denominator in the NPV formula disproportionately penalizes cash flows occurring in years 3, 4, and 5.
Following the methodology popularized by academics like Aswath Damodaran, firms must view the cost of capital as a dynamic variable that changes with market conditions, not a static internal constant Aswath Damodaran: cost of capital by industry. If your 2025 AI roadmap was approved when the 10-year was at 3.5%, a shift to 5.3% significantly reduces the present value of those future efficiency gains.
The decision framework
Management should not respond to rising rates with a "blanket cut" to all AI spending. Instead, adopt a tiered approach that distinguishes between "speed-to-value" automation and "foundational" infrastructure.
Tiered Hurdle Strategy
Project Type
Example
Payback Horizon
Recommended Review Action
Tactical / Automation
Customer service chatbot, automated document processing.
< 18 months
Maintain/Expand. NPV sensitivity to rates is low due to immediate cash inflows.
Re-run. Check if the NPV remains positive at the new WACC + a 1% risk buffer.
Foundational / Transformational
Proprietary LLM training, enterprise data fabric, core R&D.
> 36 months
Re-scope or Stage-gate. Shift from massive upfront CapEx to OpEx/usage models to de-risk.
Decision Rules
Apply Project-Specific Risk Premiums: Do not use a single corporate WACC for all AI. A "black box" AI model used in financial forecasting carries a higher risk premium than an AI tool used for internal IT ticketing.
Embrace Real-Options Thinking: Instead of approving a $50M data platform today, approve $5M for a "pilot and pivot" phase. This gives management the option (but not the obligation) to deploy more capital once the rate environment or the technology matures.
Favor OpEx over CapEx: When interest rates are high, the cost of financing large upfront hardware/infrastructure buys increases. Negotiate "pay-as-you-go" or usage-based pricing with cloud and AI providers to shift the cost of capital burden to the vendor where possible.
Mandate Utilization Milestones: Projects should not receive full funding at inception. Link subsequent funding tranches to hitting specific technical or cash-saving milestones.
The "Strategic Exception" Rule: Define a clear category for "Must-Do" projects. These are AI initiatives required for regulatory compliance or those that prevent existential competitive threats. These projects may bypass the standard hurdle rate but must still be documented for their strategic value.
Worked example
To illustrate the impact of the rate hike, consider two hypothetical AI projects under two different economic scenarios.
Note: The following values are illustrative assumptions used for demonstration purposes and do not represent actual financial forecasts.
Annual Savings: $1,200,000 (realized at end of year 1)
Old Hurdle (9%): NPV ≈ $100,917
New Hurdle (11%): NPV ≈ $81,081
Impact of 6-month delay in benefits: If savings are delayed to Month 18, the NPV at the 11% hurdle drops to approximately +$26,000, cutting its value by about two-thirds.
Scenario B: 4-Year Data Platform Build (Foundational)
Initial Outlay: $10,000,000
Annual Benefits: $4,000,000 (starting Year 2, through Year 5)
Old Hurdle (9%): NPV ≈ $1,890,000
New Hurdle (11%): NPV ≈ $1,180,000
Impact of 6-month delay in benefits: A delay in the realization of the Year 2 benefits can swing the NPV down by about $165,000 due to the high discount rate applied to those later cash flows.
Comparison Table (Illustrative)
Project
Metric
Old Hurdle (9%)
New Hurdle (11%)
Sensitivity to Delay
Workflow Automation
NPV
$100,917
$81,081
High (NPV falls about two-thirds)
Data Platform
NPV
$1,890,000
$1,180,000
Moderate (Significant value erosion)
Key Takeaway: The automation project is more "fragile" regarding timing. A small delay in implementation can erase about two-thirds of its value. The platform project remains profitable but loses about 38% of its perceived value simply because the cost of money has increased.
What to do in the next 30 / 90 days
The CFO office should execute the following plan to align the 2027 budget cycle with the current macro reality.
The 30-Day "Immediate Audit" Phase
Update WACC Inputs: Replace the 2025 "low-rate" risk-free rate with the current 10-year Treasury yield. Update the equity risk premium based on recent volatility.
Re-run the Bottom Quartile: Identify all AI projects currently in the "pipeline" or "early execution" stages that have the lowest projected IRRs (Internal Rates of Return). Re-calculate their NPVs using the new hurdle rate.
Identify "Zombie" Projects: Flag projects that were approved in 2025 but, under the new math, no longer meet the minimum hurdle. These projects should be candidates for immediate pausing or re-scoping.
Draft the Memo: Prepare a brief for the CEO and Board explaining why certain project priorities are shifting. Frame this as "capital discipline" rather than "technology retreat."
The 90-Day "Structural Reset" Phase
Rewrite the Approval Template: Update the standard capital expenditure request form. Require project leads to include a "Sensitivity to Delay" analysis and a "WACC Sensitivity" table.
Implement Stage-Gating: Formally integrate "Real-Options" into the procurement policy. Move from single-stage massive approvals to multi-stage, milestone-based funding.
Vendor Negotiation Strategy: Direct the procurement team to prioritize SaaS and OpEx-based AI contracts over heavy CapEx/on-premise infrastructure builds to preserve cash and reduce interest rate sensitivity.
Finalize 2027 Plan: Integrate the re-ranked project list into the formal 2027 budget, ensuring that capital is disproportionately allocated to high-certainty, short-payback wins and mission-critical strategic plays.
Signals to watch and revisit triggers
The decision to maintain a high hurdle rate is not permanent. Management should monitor the following signals to trigger a "re-calibration" of the hurdle rates:
Treasury Auctions and Term Premium: Watch for stability in 10-year and 30-year auction results. If the term premium (the extra yield investors demand for holding long-term debt) begins to compress, it may signal a period of lower long-term discount rates.
The 2-10 Yield Curve: Monitor the spread between the 2-year and 10-year Treasury yields. A return to a healthy upward slope (from inversion or flattening) suggests a more predictable long-term interest rate environment for planning.
FOMC Communications: While the 10-year is driven by inflation and term premiums, the Fed's "dot plot" and subsequent press conferences provide the anchor for the short end of the curve.
Credit Spreads: If corporate credit spreads widen significantly, it indicates that even if Treasury rates drop, the marginal cost of debt for your specific company may remain high.
Inflation Prints (CPI/PCE): Persistent inflation will keep the "real" cost of capital high. A cooling trend is a prerequisite for lowering hurdle rates.
Risks and what this file does not tell you
This framework provides a disciplined approach to capital allocation, but it is not without risks:
The "Wait-and-See" Risk: If you raise hurdle rates and interest rates subsequently fall, you may have paused critical AI investments that your competitors (who maintained their pace) have already integrated into their core operations. This is a "lost opportunity" risk.
Spot vs. Normalized Yields: Using today's specific spot yield to set a 5-year project hurdle can be dangerous. It is often better to use a "normalized" view (e.g., a moving average of the 10-year yield) to avoid making decisions based on temporary market spikes.
False Precision: Calculating NPV to the fourth decimal point provides a false sense of security. The underlying assumptions—such as the exact dollar amount of AI-driven savings—are often estimates. The hurdle rate adjustment is a tool for relative ranking, not for determining the absolute "correct" value of a project.
Strategic Value Blindness: This file focuses on financial metrics. It does not account for "unquantifiable" strategic value, such as the brand prestige of being an "AI-first" company or the talent acquisition benefits of working on cutting-edge platforms. These projects should be evaluated through a separate strategic lens, not just a WACC-based lens.
AI