nika

swe at @Visa | game dev
co-founder @TrueGambit_ 100M+ Visits | 2x exits

[Taylor @taycaldwell]

Announcing the 𝕏 MCP. 🖤

Use the 𝕏 API with any MCP-compatible AI tool using your own account’s permissions.

[Kalshi @Kalshi]

BREAKING: Elon Musk has made more money in the last 24 hours than Warren Buffett in his entire lifetime

[Carlos That Notices Things @QuetzalPhoenix]

The richest guy on Earth SHOULD be the guy making cutting edge cars and rockets instead of dudes who sell purses and perfumes or dudes who run investment firms or dudes who made Facebook.

[Wall St Engine @wallstengine]

Elon Musk just became the world’s first trillionaire.

Roughly 3.5% of U.S. GDP.

His SpaceX stake alone is worth roughly $866B.

That is larger than the entire net worth of almost every billionaire on Earth.

If Musk were a public company, he would rank the 13th-largest company

[R A W S A L E R T S @rawsalerts]

🚨 #BREAKING: The average American is now closer to Jeff Bezos in net worth than Jeff Bezos is to Elon Musk.

[Lance Martin @RLanceMartin]

https://t.co/es0JQM4MS9

[vivek @itsreallyvivek]

https://t.co/ZhOgyq7Vgn

[Shivam Misra @s_misra2000]

@claudeai let the monthly migration from codex to claude code begin

[Saso Capital @saso_capital]

My favorite 800V Power Semiconductor stocks ranked:

1. $WOLF (Wolfspeed) $2.7B market cap — The most asymmetric setup in the entire space. Wolfspeed controls the SiC substrate bottleneck, the foundational material every other SiC device maker needs. If the fab reaches target utilization, revenue could 3-5x from the current ~$713M run rate, and gross margins would inflect from negative to 40%+. At $2.7B, the market is pricing in heavy skepticism. A successful execution would make this a $10B+ company.

2. $NVTS (Navitas Semiconductor) $6B market cap — If Navitas captures even 2-3% of the AI data center power conversion TAM over the next three years, that’s a $500M+ revenue business on a fabless cost structure with 40%+ gross margins. The GaN IC technology is proven, the GeneSiC SiC portfolio adds a second vector, and the 59% revenue concentration in one AI-infrastructure distributor tells you where the growth is coming from. From a 3-year view, the revenue base could be 10x larger than today.

3. $AEHR (Aehr Test Systems) $1.5B market cap — The purest picks-and-shovels play on SiC scaling. Every SiC wafer that Wolfspeed, onsemi, STM, Infineon, or Rohm produces needs burn-in testing, and Aehr’s FOX-XP platform is the standard tool. Revenue is ~$100M and growing rapidly, but SiC production volumes are projected to 5-10x over the next five years. Aehr’s revenue is directly levered to that ramp with high incremental margins. A $100M revenue equipment company serving a market that’s about to quintuple has obvious math.

4. $VICR (Vicor Corporation) $5B market cap — Vicor’s factorized power architecture and power-on-package modules solve a physics problem that gets worse with every GPU generation: delivering 1,000+ amps at sub-1V to a processor die without unacceptable voltage droop and heat. If this architecture becomes standard in AI accelerator designs, Vicor’s content per server could be $500-1,000+. Revenue is ~$400M today. In a bull case where 2-3 major AI chip platforms adopt Vicor’s approach, this is a $2B+ revenue company at 50%+ gross margins, a $25B+ market cap.

5. $AOSL (Alpha & Omega Semiconductor) $1.5B market cap — The most overlooked name in power semi. AOSL makes power MOSFETs and PMICs for computing, server, and industrial applications. At $1.5B market cap and ~$650M revenue, it trades at ~2.3x sales, a fraction of MPWR’s 37x. The company is modestly profitable, has real products shipping into server power applications, and would benefit enormously from any broadening of AI power demand beyond the top-tier suppliers. This is the deep value pick: if the AI power theme lifts all boats, AOSL re-rates from 2x to 5x sales and the stock doubles without heroic assumptions.

Honorable mentions:
$DIOD $XFAB $ON $POWI $MPWR

[Wall St Engine @wallstengine]

$NVDA and SK Hynix announced a multi-year technology partnership for next-gen memory aligned with NVIDIA’s AI infrastructure roadmap.

SK Hynix will co-develop memory for Vera Rubin AI supercomputers, Vera CPUs, RTX Spark PCs, and Jetson Thor robotics platforms.

The partnership also includes using NVIDIA CUDA-X and PhysicsNeMo to accelerate chip design and simulations, plus Omniverse digital twins for autonomous fab operations.

[nisedo @nisedo_]

A must read

https://t.co/d4tiRoEvpI

[Shaan Puri @ShaanVP]

Turns out black mirror was a very good source for startup ideas.

[Rohit Ghumare @ghumare64]

As an AI Engineer. Please learn:

Harness engineering, not just prompt engineering
Context engineering, not just long prompts
Prompt caching vs. semantic caching tradeoffs
KV cache management, eviction, reuse, and memory pressure at scale
Prefill vs. decode latency and why they optimize differently
Continuous batching, paged attention, and throughput optimization
Speculative decoding vs. quantization vs. distillation tradeoffs
INT8, INT4, FP8, AWQ, GPTQ, and when quantization hurts quality
Structured output failures, schema validation, repair loops, and fallback chains
Function calling reliability, tool contracts, argument validation, and idempotency
Agent guardrails, loop budgets, tool budgets, and termination conditions
Model routing, graceful fallback logic, and degraded-mode UX
RAG architecture: chunking, embeddings, hybrid search, reranking, and freshness
Retrieval evals: recall, precision, grounding, attribution, and citation quality
Evals: golden sets, regression tests, adversarial tests, LLM-as-judge, and human evals
LLM observability as a first-class discipline: traces, spans, tokens, latency, errors, and drift
Cost attribution per feature, workflow, tenant, and user journey not just per model
Safety engineering: prompt injection defense, data leakage prevention, and permission boundaries
Multi-tenant isolation, cache safety, and cross-user context contamination prevention
Fine-tuning vs. in-context learning vs. RAG vs. distillation and when each is the wrong tool
Latency, quality, cost, and reliability tradeoffs across the full inference stack
Production failure modes: hallucinated tool calls, malformed JSON, stale retrieval, runaway agents, and silent eval regressions
Shipping LLM systems as reliable infrastructure, not demos wrapped around prompts

https://t.co/OhK9MK04ld

[Alps @alpaysh]

i still think about this

[Ian Tracey @ian_dot_so]

Things are going to get very wierd very soon

[27. @dimvji]

idk how to explain it but i have just two years to figure it out

[Xeusthegreat (♟,♟) @SamuelXeus]

Banks turned $1,000 into $600,000 in just one year.

Let me explain what actually happened and how you can position yourself to do the same thing.

First, you need to understand what an ICO is.

ICO stands for Initial Coin Offering. Think of it like a company selling shares before it goes public on the stock market, except instead of shares, they are selling tokens.

When a crypto project wants to raise money to build their product, they sell their token early at a low price. The people who buy in early get in cheap and if the project does well and the token goes up in value, those early investors make a lot of money. That is the basic idea.

The problem for years was that only VCs and wealthy insiders got access to these early deals. By the time regular people heard about a project, the price had already gone up and the insiders were already in profit. Platforms like Echo and Legion exist to fix that.

Now let me show you what is possible when you get access.

LAB Terminal that returned over 600x for @Mrbankstips was on Legion. Another one called Sonar's XPL token sale hit a 33.78x return at its all-time high.

To put that in simple terms, someone who put in $1,000 on the ICO walked away with over $600,000. That is what early access looks like in practice.

Here are some of the platforms where you can get early access to rounds

1) @echodotxyz

Echo works like a group investment club. A lead investor who has done their research says I found a good deal and opens a pool. You follow them, put in your money alongside theirs, and everyone gets into the deal at the same early price.

2) @legiondotcc

Legion works differently. Instead of following someone into a deal, you build your own reputation on the platform and that reputation determines how much access you get.

The platform looks at how active you are in the crypto space, your on-chain history, your participation in the ecosystem, and gives you a score. The better your score, the better your allocation.

Both platforms are doing the same thing. Giving regular people access to deals that used to belong only to insiders.

How to spot good projects on these platforms:

• Look at who is leading the investment. If credible and experienced people are putting their money in, that means something
• Research the team. Have they built anything before? Do they have real experience?
• Check the valuation at launch. If the fully diluted value is very high compared to what is actually circulating, be careful
• See who else is investing alongside you. Serious names in a round change the risk profile
• Check how long the team's tokens are locked up. If they can sell immediately after launch, that is a red flag
• Look for proof that people are already using the product before the token exists
• Do not rely on one source of information. Always dig deeper

The opportunity is real. But getting access without knowing what you are doing is just an expensive way to learn a lesson.

Learn first. Then move.

As you were. ❤️🦅

[MEH @InvestwithMEH]

Anthropic 🤝 @RobinhoodApp

Welcome to the $HOOD Claude.

[kache @yacineMTB]

We are in a golden age where, if you are good at systems and understanding, AI increases your abilities by an order of magnitude. But if you are not good at it, you just spin your wheels and end up nowhere helpful