Hewlett-Packard's Merced Decision Custom Case Solution & Analysis

1. Evidence Brief (Case Researcher)

Financial Metrics

HP R&D Budget (1998): $2.2 billion.
Merced (IA-64) Development Cost: Estimated $500 million to $1 billion (Para 4).
HP Operating Profit (1997): $3.1 billion on $42.9 billion revenue (Exhibit 1).
RISC (PA-RISC) R&D spend: Significant portion of annual R&D budget (Para 12).

Operational Facts

Merced: Joint development project between HP and Intel for 64-bit architecture (Para 1).
Transition: Moving from PA-RISC proprietary architecture to Intel-standard IA-64 (Para 7).
Timeline: Merced delivery delayed; originally expected 1997, pushed to 1999/2000 (Para 18).
Market Position: HP holds 30-40% of the Unix server market; critical for high-margin revenue (Exhibit 2).

Stakeholder Positions

Lew Platt (CEO): Focused on industry-standard transition to maintain long-term relevance.
Intel: Pushing for industry-wide adoption of IA-64 to commoditize the server space.
HP Engineering/R&D: Concerned about losing differentiation and the performance gap between PA-RISC and early IA-64 iterations.

Information Gaps

Exact cannibalization rate of PA-RISC by IA-64 is not quantified.
Specific cost-per-unit differential between internal RISC chip manufacturing and Intel sourcing.

2. Strategic Analysis (Strategic Analyst)

Core Strategic Question

Should HP abandon proprietary PA-RISC development to commit fully to the Intel IA-64 architecture, despite performance risks and timing delays?

Structural Analysis

Value Chain: HP currently controls the stack (chip and system). Moving to Intel shifts value to the chip supplier, forcing HP to compete solely on system-level integration.
Ansoff Matrix: This is a product development gamble. HP is betting that the market will shift to a standardized 64-bit architecture faster than internal proprietary performance can sustain.

Strategic Options

Option 1: Full Migration to IA-64. Stop all PA-RISC investment. Trade-off: Total reliance on Intel; risk of performance lag; potential alienation of existing Unix customers.
Option 2: Parallel Development. Maintain PA-RISC for high-end users while phasing in IA-64 for mid-range. Trade-off: Doubled R&D costs; internal resource fragmentation; slower market penetration of IA-64.
Option 3: Selective Licensing. License PA-RISC to third parties to keep the architecture alive while transitioning. Trade-off: Loss of exclusivity; potential loss of control over the platform roadmap.

Preliminary Recommendation

Pursue Option 1. Proprietary chip development is a declining moat. The shift to industry-standard hardware is inevitable. HP must pivot to software and system integration to maintain margins.

3. Implementation Roadmap (Implementation Specialist)

Critical Path

Software Portability Layer: Ensure binary compatibility for legacy PA-RISC applications on IA-64 before the hardware launch.
Customer Migration Incentives: Develop a three-year transition support program for existing Unix accounts.
R&D Reallocation: Immediately redirect 40% of PA-RISC hardware engineers to IA-64 platform optimization.

Key Constraints

Intel Execution: HP is hostage to Intel’s release schedule. Delays are outside HP’s control.
Performance Parity: If Merced underperforms compared to legacy RISC chips, high-end customers will defect to IBM or Sun Microsystems.

Risk-Adjusted Strategy

Maintain a skeleton crew for PA-RISC maintenance for 24 months post-launch. Do not initiate new silicon designs. If Intel misses the 1999 window, immediately extend PA-RISC support for 18 months to prevent customer churn.

4. Executive Review and BLUF (Executive Critic)

BLUF

HP must commit to IA-64. The company is trying to solve a hardware problem with a software reality: proprietary silicon is a depreciating asset. The threat is not Intel—it is the erosion of the Unix server market by lower-cost, standard-based competitors. PA-RISC has served its purpose; further investment is a sunk-cost fallacy. HP must transition its engineering focus from chip architecture to high-value software and enterprise integration. The risk of waiting is the total loss of the Unix base to firms that embrace the industry standard faster.

Dangerous Assumption

The analysis assumes Intel will deliver a competitive product on time. Given historical delays, this is a fatal premise. If Intel slips, HP has no fallback plan.

Unaddressed Risks

Competitive Defection: Sun Microsystems will aggressively target HP’s high-end Unix base during the transition period.
Margin Compression: Moving to Intel standardizes the hardware, likely lowering HP’s pricing power compared to the proprietary era.

Unconsidered Alternative

The team failed to consider a targeted acquisition of a software-focused firm to bolster the application stack, which would allow HP to differentiate through software even as hardware becomes a commodity.

Verdict: APPROVED FOR LEADERSHIP REVIEW.