Search on TFTC
Is AI CAPEX the Next Electricity Grid or the Next Dark Fiber? BTC Drops Below $65K as Fear Hits Historic Lows

Is AI CAPEX the Next Electricity Grid or the Next Dark Fiber? BTC Drops Below $65K as Fear Hits Historic Lows

Feb 24, 2026
Bitcoin Brief

Is AI CAPEX the Next Electricity Grid or the Next Dark Fiber? BTC Drops Below $65K as Fear Hits Historic Lows

TFTC – Truth for the Commoner

Bitcoin Brief

Sup, freaks.

The largest companies on earth are spending $610 billion on AI infrastructure this year. The question nobody can answer is whether this is the electricity grid of 1900 or the dark fiber of 1999. History says you cannot tell the difference in real time, and the consequences of getting it wrong are enormous. Meanwhile, Bitcoin dropped below $65,000, the Fear and Greed Index hit 5 (the lowest reading in its history), and half a billion dollars in leveraged positions were wiped out in 48 hours. The market is telling you something. Whether you listen is up to you.


LEAD STORY

$610 Billion Question: Is AI CAPEX the Next Electricity Grid or the Next Dark Fiber?

The numbers are staggering. Microsoft, Google, Meta, Amazon, and a growing list of hyperscalers have committed roughly $610 billion in combined AI capital expenditure for this year alone. Nvidia cannot ship GPUs fast enough. SK Hynix and Samsung are sold out of high-bandwidth memory through 2026. Sony has reportedly pushed the PlayStation 6 to 2028 or 2029 because they cannot secure fab capacity at TSMC. Elon Musk is floating the idea of building a dedicated semiconductor fab for xAI. The AI infrastructure buildout is the largest coordinated capital deployment in the history of the technology sector.

The optimistic analogy is electrification. When Edison and Westinghouse built out the power grid in the late 1800s, the upfront costs were enormous and the skeptics were loud. But electricity turned out to be a general-purpose technology that reshaped every industry it touched. The pessimistic analogy is dark fiber. In the late 1990s, telecom companies laid millions of miles of fiber optic cable on the assumption that internet bandwidth demand would grow exponentially forever. They were right about the trend and catastrophically wrong about the timing. Companies like WorldCom, Global Crossing, and 360networks went bankrupt. The fiber eventually got used, but the investors who funded the buildout were wiped out first.

The uncomfortable truth is that both analogies are partially correct, and you cannot tell which one dominates until after the fact. As Citrini Research pointed out, the core problem is the gap between infrastructure investment and revenue generation. ChatGPT's share of "General Research" usage has flatlined at 36-37% for months. Enterprise adoption is growing but remains heavily concentrated in a few use cases. The revenue to justify $610 billion in annual spend simply does not exist yet. That does not mean it will never exist. It means the market is pricing in a future that has not arrived, which is exactly what happened with fiber.

The dark fiber parallel runs deeper than most people realize. In 1998, telecom analysts had models showing bandwidth demand doubling every 100 days. The models were not wrong about the direction. Internet traffic did explode. But the timing was off by years, and the capital markets did not have the patience to wait. WorldCom fabricated $11 billion in earnings to paper over the gap between infrastructure spending and revenue. Global Crossing burned through $30 billion. The fiber they laid is still in the ground, and it eventually became the backbone of the modern internet. But the equity investors who funded the buildout got zero. The technology was transformative. The investment thesis was a catastrophe.

Today's AI CAPEX has the same structural problem. The infrastructure is being built on the assumption that AI revenue will materialize fast enough to justify the spend. But the history of general-purpose technologies says that adoption curves are S-shaped, not exponential. Electricity took 30 years to reach 50% of American factories after the first commercial power plant opened. The internet took 15 years from ARPANET to commercial viability. AI may compress that timeline, but even a five-year gap between infrastructure buildout and revenue realization is enough to destroy the shareholders who funded it.

NVDA reports earnings Wednesday. If guidance disappoints, the dark fiber narrative takes hold fast. If it beats, the electricity narrative gets another quarter of oxygen. Either way, $610 billion is a lot of chips to push into the middle of the table on a thesis that is still unproven. The question is not whether AI will be transformative. It is whether the investors funding the buildout will survive long enough to see the returns.


SIGNAL

Jane Street Accused of Insider Trading That Helped Collapse Terraform

$150M redeemed through back channels. $85M shorted on Curve. A secret chat group called "Bryce\'s Secret."

The Wall Street Journal reports that Terraform Labs is suing Jane Street, one of the most powerful trading firms on Wall Street, alleging insider trading that helped trigger the catastrophic collapse of the TerraUSD (UST) stablecoin in May 2022. At the center of the lawsuit is a private Signal chat group called "Bryce's Secret," where traders from Jane Street and other firms allegedly shared confidential information about Terraform's internal redemption mechanisms.

According to the complaint, Jane Street redeemed approximately $150 million in UST through a privileged back-channel mechanism that was not available to ordinary holders, effectively front-running the depeg. The firm then allegedly opened an $85 million short position via Curve Finance as UST began its death spiral, profiting from the very collapse it helped accelerate. The depeg ultimately wiped out roughly $40 billion in value across the Terra ecosystem and sent shockwaves through the broader crypto market, contributing to the cascade of failures that took down Three Arrows Capital, Celsius, and Voyager.

If the allegations hold up, it would confirm what many in the bitcoin community have long suspected: the TradFi giants who lecture about "market integrity" and lobby for tighter crypto regulation are the same ones exploiting information asymmetries behind closed doors. Jane Street is not some fringe operation. They are one of the largest market makers in the world, handling roughly 5% of all U.S. equity trading volume. The idea that a firm of that caliber was allegedly colluding in a secret chat to front-run a stablecoin collapse should make everyone reconsider who the real bad actors in this space are.

Bitcoin Drops Below $65K as Fear Hits Historic Lows

Fear and Greed Index at 5. Half a billion liquidated. The market is capitulating.

Bitcoin fell below $65,000 over the weekend, extending a drawdown that has now wiped nearly 40% from the all-time high. The Fear and Greed Index hit 5, the lowest reading in the index's history, deeper than the FTX collapse and the Terra/Luna implosion. Over the past 48 hours, more than $500 million in leveraged positions were liquidated across exchanges, with longs accounting for 93% of the damage. A single $61.5 million long on HTX was the largest individual liquidation. The daily RSI sits around 30, deep in oversold territory. Short-term holders are 27% underwater relative to their cost basis ($89,119 realized price vs. $64,707 spot). Net realized losses hit $105 million per day. This is what capitulation looks like on-chain: sellers exhausting themselves, weak hands transferring coins to stronger ones. James Check of Checkonchain published a timely piece noting that Bitcoin is at the same price as May 2024, when he first coined the term "chopsolidation" to describe extended sideways digestion after major moves. His framework: markets don't travel in straight lines. Major moves require either prolonged sideways consolidation or counter-trend corrections before the next leg. He correctly called the months-long chop after the $73K ATH in March 2024. The question now is whether the current drawdown is another chopsolidation phase that resolves higher, or the start of something worse. Historically, Fear and Greed readings below 10 have preceded significant reversals, but "historically" is doing a lot of heavy lifting in a macro environment this uncertain.

The KYC Industry Just Exposed 1 Billion People

A KYC company left a billion records in an unprotected database. No password.

Cybernews reported that IDMerit, a KYC and age verification provider used by businesses worldwide, left an unsecured MongoDB database exposed with approximately one billion personal records from 26 countries. Names, home addresses, phone numbers, national ID numbers, dates of birth. 204 million records from the United States alone. IDMerit is the backend provider that other companies outsource identity verification to. You never consented to them having your data. You probably didn't know they existed. Every KYC regulation creates a honeypot. Every age verification law creates another. The more data you force companies to collect, the more data there is to steal. This is why cryptographic proof without disclosure matters. The tools to prove identity conditions without handing over your life story to a passwordless MongoDB instance exist today. The political will to adopt them does not.

Bitcoin Payments as Easy as Apple Pay

Tap. Done. Zero platform fees.

Numopay just launched an open-source point-of-sale app that turns any NFC-enabled Android phone into a Bitcoin payment terminal. The merchant experience mimics Apple Pay: customer taps their phone, payment settles instantly. Under the hood, it uses Chaumian eCash (Cashu) for the tap-to-pay layer and falls back to Lightning for wallets that don't support ecash. No platform fees (compare that to Visa's 2.9% + 30 cents), no chargebacks, no three-day settlement delays. It works offline too, with payments syncing when connectivity returns. BTCPay Server integration is coming. This is what merchant adoption actually looks like, not corporate partnerships, but open-source tools that anyone can run.

UK Rape Gang Inquiry Issues First Statement

The investigation that the British establishment tried to bury is now official.

MP Rupert Lowe shared the first official statement from the UK Rape Gang Inquiry, launched to investigate the systemic grooming and abuse of children across multiple British cities. The tweet pulled over 10 million views in hours. For years, local authorities, police, and social workers were accused of ignoring or actively covering up organized abuse networks for fear of being labeled racist. This inquiry represents a political inflection point in the UK. Whatever your politics, the core issue is institutional failure: government agencies chose optics over the safety of children. The lesson for any system, financial, political, or social, is that centralized institutions will always prioritize self-preservation over the people they claim to protect.

Anthropic Catches DeepSeek, Moonshot, and MiniMax Stealing Claude's Brain

24,000 fake accounts. 16 million exchanges. Industrial-scale AI theft.

Anthropic published evidence that three Chinese-linked AI labs, DeepSeek, Moonshot AI, and MiniMax, ran industrial-scale "distillation attacks" against Claude. They created over 24,000 fraudulent accounts and generated 16 million exchanges to extract Claude's capabilities and train their own models. The technique is straightforward: feed a stronger model millions of prompts, collect its outputs, and use that data to train a cheaper model that mimics the original. Anthropic frames this as a national security threat, arguing that distilled models strip out safety guardrails, meaning dangerous capabilities proliferate without protections. Elon Musk responded by pointing out that Anthropic itself has paid multi-billion dollar settlements for training data theft. The irony is thick. The broader takeaway: AI model weights are the new trade secret, and nobody has figured out how to protect them. The "rapid advances" of open-source Chinese models may owe more to American R&D than anyone wants to admit, which actually strengthens the case for export controls on advanced chips.

Bitcoin OpTech: OP_RETURN Data Is In, PIPEs v2 Enables Covenants Today

24 million OP_RETURN transactions, and the sky didn't fall. Plus covenants without a soft fork.

Two highlights from Bitcoin Optech Newsletter #393. First, Anthony Towns published data on OP_RETURN usage since Bitcoin Core v30.0 relaxed policy limits in October. Of 24.3 million transactions with OP_RETURN outputs, only 61 had multiple outputs and only 396 exceeded the old 83-byte limit. Large outputs accounted for just 0.44% of total OP_RETURN data. The debate was louder than the impact. Second, Misha Komarov published Bitcoin PIPEs v2, a protocol that achieves covenant-like spending conditions without any consensus changes. The trick: encrypt a private key behind a predetermined condition using witness encryption. If the condition is met, the key is revealed and a valid Schnorr signature can be produced. All conditional logic stays off-chain. This gives developers binary covenant capabilities today, without waiting for a soft fork.


DATA SNAPSHOT

Bitcoin Price$64,707
Sats per Dollar1,545
Block Height938,033
Network Hashrate1,019 EH/s
Priority Fee1 sat/vB

On-Chain Metrics
Fear & Greed Index5 Extreme Fear (historic low)
Daily RSI~30 Deep oversold territory
48h Liquidations$508M 93% longs, 146K+ traders
MVRV Ratio1.24 Fair value range
STH Realized Price$89,119 Short-term holders 27% underwater
SOPR0.995 Coins moving at a loss
NUPL0.19 Hope/Fear zone
Net Realized P/L-$105M/day Sell pressure fading

If this landed, forward it to someone who could use more signal and less noise. The Bitcoin Brief is free, always will be.

See you tomorrow,

Marty Bent


Follow: @MartyBent · @TFTC21

Nostr: primal.net/marty

YouTube: TFTC · Podcast: tftc.io/podcast

Spread the signal,
earn Bitcoin.

Get your unique referral link when you subscribe.

Current
Price

Current Block Height

Current Mempool Size

Current Difficulty

Subscribe