Enterprise AI

Why Serve Open-Source LLMs with vLLM on Internal GPUs?

Deciding internal open-model serving for security, data control, cost, and performance

When internal GPU serving and vLLM make sense: compare external APIs, adoption trade-offs, batching and prefix caching in plain terms, and a pre-flight checklist.

Bloom Energy: An AI Power Bottleneck Winner, but How Much Success Is Already Priced In?

A closer look at the Oracle agreement, SOFC economics, competitive moats, and the valuation implied by Bloom Energy’s AI power boom.

Bloom Energy’s AI-power opportunity is real. We test contracts, cash flow, dilution, and the value that can reach common shareholders.

Applied Digital Corporation: $36 Billion in AI Data Center Leases, but How Much Value Reaches Shareholders?

1.41GW of contracted capacity, an 89% NOI margin, and billions of dollars of construction spending.

A contract-to-common-equity analysis of Applied Digital Corporation's AI data-center portfolio.

Salesforce, Inc.: Can Agentforce Reaccelerate CRM Growth?

AI revenue is growing fast. The real question is whether it survives cannibalization and becomes free cash flow per share.

A Salesforce reverse DCF testing organic growth, cRPO, Agentforce economics, SBC, buybacks, and the FCF growth embedded in today’s CRM price.

From Retrieval Engineering to Task Engineering

What changes when retrieval becomes part of an end-to-end task?

AI engineering is moving from optimizing retrieval pipelines to specifying, constraining, and evaluating end-to-end enterprise tasks.

Will Rogo Eat the Financial Intelligence Stack? Two Years with AlphaSense Changed My Answer

A financial AI market map that separates banking value creation from vendor revenue and a seat-based SAM, then asks where AlphaSense and Rogo can build lasting moats.

AWS Is Hiring FDEs. But Can You Hire Your Way Into an FDE Organization?

What hyperscalers and Palantir reveal about customer-to-product loops and accumulated learning

A framework for evaluating whether hyperscaler FDE teams can turn field discoveries into reusable products, with Palantir as the counterexample that makes the thesis testable.

FDE Economics: What Wonderful and Reflection Need to Prove

Field engineering becomes software economics only when today’s deployment makes the next one cheaper and faster

A private-market analysis of Wonderful's labor leverage and Reflection's challenge of turning open intelligence into capital and distribution leverage.

Who Disappears First: The AI Team or the Business Teams?

As agents take on more work, which team changes first? A task-level view of how enterprise AI may reshape both builders and business practitioners.

Why Enterprise AI Transformation Should Start Narrow

General-purpose AI is a Swiss Army knife; enterprise AX is a purpose-built blade. A Market Intelligence case shows why transformation means redesigning work, not automating it.

Applied Digital Corporation Valuation Lab: What Is 1.41GW of Contracted Capacity Worth Per Share?

Change NOI, residual CAPEX, discount rate, re-letting value, and incremental MW to test the additional NAV implied by the market.

An interactive sensitivity tool connecting Applied Digital Corporation's contracted capacity, development cost, and capital claims to common-equity value.

Salesforce Valuation Lab: The FCF Growth Implied by Today’s Price

Move FCF growth, the SBC adjustment, and the discount rate, then reverse-solve the growth embedded in the market price.

Change Salesforce FCF growth, the SBC adjustment, the discount rate, and terminal growth to calculate five-year fair value and the growth implied by today’s price.

AI Handoff Queue Simulator

How do handoff rate, handling time, and staffing change customer waits?

Change an AI contact center's human handoff rate, average handle time, and staffing to compare wait times and service levels in an M/M/c queue.

Why Is vLLM Slow?

Diagnosing bottlenecks with TTFT, TPOT, and KV cache

A practical guide to diagnosing vLLM serving bottlenecks by reading TTFT, TPOT, queue, prefill, KV-cache, and preemption signals together.

What is vLLM tuning?

Serving LLMs when GPUs are scarce

A practical guide to serving LLMs on limited on-premises GPUs: diagnose prefill, decode, KV cache, queues, and tail latency to tune vLLM for sustainable throughput.

Build with Gemini Notes: Google Cloud Is Taking the Agent Era Seriously

How the number-three cloud is using agents as its next enterprise entry point

Notes from Build with Gemini on Google Cloud's agent governance, Gemini Enterprise as a sticky entry point, and its distribution challenge in Korea.

Why Our Agent Platform Got Off to a Rocky Start

From a convincing demo to a predictable production system

Our agent platform worked in demos but stumbled in production. A retrospective on predictability, trust debt, and earning abstractions through deployment.