Trend Judgment · Product Practice · Thinking Notes

Phoenix Lee

Turn unclear trends into product direction.

Locsic.com is my independent research and writing space for AI infrastructure, agent systems, and the future of software. I turn technical signals and real problems into product judgment, hands-on experiments, and long-form analysis, keeping the evidence and trade-offs visible.

Recent thinking

31 min read

Same Node, Double the Compute: How Ascend 960 Rebuilds the Chip, the Supernode, and the Cluster

With the process node pinned at N+2, Ascend 960 doubles compute through three system-level levers: chiplet packaging and in-house HBM at the card level, a 4,096-card supernode whose 5,500 NPO optical engines replace 48,000 pluggable modules, and cluster topology that fences communication inside a 2-microsecond latency domain.

Full summary

Set half a generation off from Rubin Ultra in 2027, the domestic inference cost curve is starting to decouple from GPU pricing.

50 min read

Fitting a Trillion Parameters into One Machine: HBF and the Three Knobs of Memory Provisioning

A September Huawei paper offers a configuration answer for single-node trillion-parameter appliances: weights in HBF, the state tier on commodity LPCAMM2 modules, and one quarter of internal bandwidth exposed to the host.

Full summary

The article verifies the knees of three independent budget lines and maps the debate between OXMIQ and five academic papers; the criticism axes differ, but on weights belonging in HBF the verdict is unanimous, and both camps marked the same land on the map.

35 min read

The Two-Trillion-Dollar Circular Ledger: Anthropic's IPO and the First Public Reckoning of AI Capital

The outline of the largest IPO in history is complete: Nvidia is in talks to anchor up to $10 billion as Anthropic seeks to raise $100 billion at a $2 trillion valuation, past the records SpaceX set in June.

Full summary

This editorial lays the past ten months of circular contracts into a ledger: behind every equity investment sits a larger compute-procurement commitment, and the money comes back once through infrastructure bills, again through equity appreciation. With SpaceX’s prospectus as the control group, three judgments: the listing documents will lay the ledger open for the first time; what this IPO prices is capital-expenditure endurance; governance and the safety narrative become the balance-sheet entries.

What I Observe

All topics →

Not every hot topic deserves attention. I track the technology, company, product, and trend signals that may change judgment, writing, or product direction.

AI Infrastructure Rebuild

Tracking AI infrastructure from chips to complete systems: supernodes, Rubin rack BOMs, UnifiedBus and AI networking, CPO optical interconnect, memory, power, cooling, and deployment economics.

Explore topic
Current read Rubin rack economics, UnifiedBus, supernodes, and CPO are converging into one systems-engineering thesis.

The Future of Software

Tracking AI coding, code-world modeling, developer tools, tests, collaboration, and delivery as software work moves toward contextual, verifiable, memory-bearing operating surfaces.

Explore topic
Current read The Cursor acquisition signal, code-world modeling, and agent work surfaces point to software tools becoming operating systems for work.

AI Reshapes Work

Tracking ResearchStudio, Loop Engineering, agent infrastructure, skills, memory, permissions, review, and rollback as AI moves from capability demos into durable research and operating workflows.

Explore topic
Current read ResearchStudio and Loop Engineering show that reusable skills, verification, and review loops are becoming the real unit of agent work.

Putting judgment into practice

View all work →
Working methods and principles

I work where answers are not yet obvious.

I do not chase every trend. I judge which changes will actually reshape resource allocation. I do not rush into solutions. I first make sure the problem is worth solving. I do not make ideas louder. I make real value visible.

Trend judgment

Find the changes worth taking seriously in technology shifts, industry developments and weak signals.

Signals → Patterns → Direction

Product practice

Turn judgment into tools, interfaces and experiments that can be examined, tested and improved.

Problem → Prototype → Workflow

Lasting ideas

Use long-form writing to preserve assumptions, trade-offs and evolving views, with context worth returning to.

Observation → Essay → Memory
Define the problem before accelerating resources.
Strategy, product, and narrative start creating commercial power when they help teams form shared conviction faster.
Explore my method →