Hot takes.

Marketing fails when it confuses big-B Brand (company soul and mission) with small-b brand (visual identity and campaigns)—a shiny logo cannot fix a weak mission.

Adding JSON-LD schema is not the AI visibility lever many hope for; a study of nearly 1,900 pages found no significant citation boost across Google or ChatGPT.

The Fed cannot pull a Volcker with $40T in debt, meaning negative real rates and quiet debt monetization are inevitable, accelerating the end of dollar hegemony.

Despite a new Fed Chairman, higher-than-expected inflation, an energy crisis, and ongoing geopolitical tensions, the S&P keeps making new all-time highs — enjoy the bull run while it lasts.

Energy markets have assumed pipes and tankers always work — that assumption has collapsed and we're now in a regime where physical scarcity, not paper trading, determines oil prices.

This isn't just a field trip — it's a tactical truce to get symbolic wins like Boeing orders and soy purchases to lower trade war temperature while the U.S. is distracted by the conflict in Iran.

The pace of development in the AI industry is overriding the typical forces of vendor stickiness, as evidenced by Anthropic overtaking OpenAI in business adoption.

AI safety research is too focused on technical abuse demonstrations rather than incident data, and the real risks from agentic systems will come from human behavior and over-delegation of control.

Wi-Fi 8 signals that future wireless upgrades will be about fixing real-world reliability pain points rather than chasing headline speed numbers.

Companies piling on copilots and AI vendors without governance risk repeating SaaS sprawl mistakes, making AI another unmanaged cost center.

PreStocks' dashboard showing an implied Anthropic valuation above $1.5 trillion despite holding only ~$23M in total assets exposes these synthetic instruments as having no recognized claim on underlying equity.

Capital is selecting for product velocity over narrative, with AI tokens decoupling from BTC correlation across two consecutive quarters and trading as proxies on the global AI buildout thesis.

If Clarity Act fails, the industry enters a roughly three-year 'show me' period where it must prove crypto is indispensable without regulatory tailwinds.

Teams that add semantic metadata to tokens and build governance frameworks will be better positioned than those optimizing only for faster component generation.

Instagram's Instants feature arrives as many users already rely on Stories for quick updates, raising questions about whether the new feature fills a genuine gap.

Dropping Face ID on a $2,000 premium foldable device may be seen as a downgrade, though the tradeoff enables a thinner, more portable form factor.

Beijing's push for AI independence, including a big bet on open-source, is allowing China to resist US pressure during trade talks.

It's time to reset expectations about AI, as a populist backlash is building against the technology.

AI is 'rotting the brains' of developers, causing them to lose their previous abilities to do their jobs.

The AI resurrection economy is growing faster than safeguards, creating uneven enforcement and significant legal and economic tension around unauthorized use of deceased artists.

Marketers should stop planning around formats like TV, social, or PR and start planning around business goals as media channels blur into one video ecosystem.

Mandating AI use across every employee role led to performative adoption without real productivity gains, and the blanket evaluation policy was wrong.

Microsoft's AI-driven direction, lack of independent leadership, and US jurisdictional data risks make GitHub an untrustworthy platform, prompting migration to self-hosted Forgejo.

Creating a custom programming language for web development was a mistake due to high developer adoption friction and the prohibitive cost of maintaining custom IDE tooling.

The Emacsification of software via AI agents will shift the future toward highly configurable, bespoke personal tools rather than generic one-size-fits-all applications.

Using AI token usage or project complexity as performance metrics creates perverse incentives that reward visible activity over actual productivity, mirroring Goodhart's Law and the Cobra Effect.

Both companies are racing toward IPOs potentially this fall, and Anthropic's enterprise data shift means it is now negotiating from a position of strength.

The complaint that Claude upsells users to pricier models is a strange criticism to level at a company gaining market share — 'they upsell too well' is a weird thing to complain about when you're losing.

The old asymmetric broadband model of fat downstream and thin upstream was engineered for passive consumption — that model is finished due to AI-driven upstream traffic.

There is no loyalty to specific AI labs — open source models are gaining traction while Anthropic leads in spend and Google leads in volume.

A protocol-led recovery from the exploit proves DeFi can move fast when it has to.

Native BTC borrowing without wrappers has been a 'holy grail' product for years, and Aave's Babylon integration could finally deliver it at scale.

Two Wall Street behemoths filing for stablecoin reserve products signals that stablecoins are definitively here to stay.

Tuesday evening's CLARITY draft continues to bring the industry closer to well-defined rules of the road, with the decentralization certification process being a meaningful step forward for crypto projects.

Preferring 'blockchain' over 'crypto' is a smart PR move given the latter has been tainted by speculations, grifts, crashes, and frauds.

Ben-Sasson's claim that blockchain mass adoption is 'inevitable' might be overstated, but the optimism is welcome and sorely needed given today's more-chastened crypto sentiment.

If the CLARITY Act fails tomorrow, crypto could return to regulation-by-enforcement and it could take years — if ever — for the bill to come back around.

The path to passage exists, but the margin for error is gone — Gillen flipped bearish on the CLARITY Act two weeks ago.

Despite the war in Iran, oil prices, and holding company consolidation, Cannes Lions 2026 shows no meaningful pullback, suggesting the festival has become too entrenched in the industry to be disrupted by macro headwinds.

As AI can synthesize existing content at scale, differentiation will increasingly depend on net-new knowledge from people actually doing the work and sharing what works.

In B2B, great taste means coloring skillfully within the lines with crayons, not painting a Van Gogh — refusing to do listicles or 'State of X' reports because they feel beneath you is costing marketers real outcomes.

Taste is currently overhyped on LinkedIn and will become a meaningless buzzword within six weeks; while valuable for angles and examples, it becomes harmful when used to justify refusing formats that platforms algorithmically reward.

As MU and SK Hynix go higher, NVIDIA-based infrastructure gets more expensive to build and operate, making the Cerebras pitch—a chip that sidesteps HBM costs entirely—easier to make to every CFO.

Sam Altman being simultaneously a personal investor in Cerebras, CEO of its largest customer, and connected as a creditor is exactly the kind of relationship structure that makes institutional investors and compliance teams nervous.

The author took a massive position in MU about a month ago and it has already gone up over 100%, as memory is becoming one of the real bottlenecks of the AI economy.

The phase of building the biggest AI brain is largely done; the inference market growing from $106B to $255B by 2030 is the whole game, and Cerebras arrived at the IPO window right as this shift accelerates.

Despite the engineering marvel and oversubscription frenzy, Cerebras trades at 91x trailing revenue, 86% of revenue is from UAE-tied entities, and the $20B OpenAI contract is conditional—making the valuation extremely risky.

AI should be used as a finishing tool, not a starting one — engage in self-driven cognitive effort first before leveraging AI assistance.

We're building smarter AI interfaces while quietly accumulating 'cognitive debt' — the irony is that better AI conversation tools may be making humans worse at thinking.

Brand approval processes mean most trend content attempts miss the boat entirely, given that 27% of consumers say trend content is only effective within 24-48 hours of a trend's lifespan.

Many social media marketers care more about social media than marketing itself, failing to develop the core skill of making content stand on its own without trends or memes.

Employee-generated content is a terrible idea for most brands, lumped in with other copycat social trends that hurt more than help.

Over-reliance on trend content dilutes brand identity, attracts non-buyers, disrupts workflow, and often misses the timing window — original content is harder but builds brands over time.

Today's LLMs predicting text alone will not lead to human-level intelligence because language is only a small fraction of how humans understand the world; future AI must use world models that learn physics, causality, and consequences.

AI code analyzers are now table-stakes — any project that hasn't run one likely has a backlog of findings waiting — but must be paired with traditional defenses.

Mythos's lighter haul on curl reflects diminishing returns on a heavily fuzzed codebase, and AI tools still only surface known bug classes — not novel ones; practitioners should discount 'dangerously good' vendor framing until independent results land.

Palantir defied every rule of B2B software growth and represents what platform shifts look like when they collide with product-market fit.

The SaaSpocalypse narrative is shifting fast as Twilio, Atlassian, Datadog, Cloudflare, and Palantir prove B2B reacceleration is real, though growth remains uneven.

Anthropic's $30B ARR in 15 months isn't a knock on Salesforce but a redefinition of what scale means in the AI era, and should recalibrate how B2B founders think about growth.

Audience research is the heart of all marketing strategy, and disciplines like ICP development and voice of customer are all just different names for the same foundational activity.

Versa's upcoming AI agent control framework is described as 'yet another sign that agent governance is quickly becoming a real enterprise security category.'

Instructure's deal with hackers to prevent data publication signals that breach response increasingly includes negotiation with attackers, not just remediation.

Measuring AI adoption by activity (token usage) incentivizes performative usage rather than real productivity gains, as seen with Amazon, Meta, and Microsoft employees gaming internal metrics.

Most '10x' claims in marketing copy don't reflect real 10x — real 10x is when the old way stops making sense entirely, like email vs fax, not '23% faster but rounded up'.

Founders who stay too secretive miss critical connections and opportunities; basic visibility like a domain or LinkedIn update signals commitment to investors and partners.

Grok is 'barely growing within enterprise organizations' and remains a rounding error despite Elon Musk's prominence.

AI-native SaaS may cap at 17% gross margins because personalization eliminates caching benefits and reasoning models burn 10-100x more tokens, making decks that assume 80% SaaS margins dangerously wrong.

Complexity spreads like a virus when not managed, and well-constructed systems require leadership that knows what to say 'No' to.

The creative community is divided — many oppose generative AI due to copyright, environmental, and job concerns, while others see value in AI for repetitive tasks but maintain that human creativity and taste remain essential and difficult to replace.

MARA's sale of $1.5B in bitcoin to fund AI infrastructure acquisition signals a meaningful strategic pivot away from pure Bitcoin mining.

OpenAI and Anthropic require the model layer to stay premium, a premise being arbitraged away as Chinese open-source models grew from 1.2% to 30% of global usage in 2025.

DeFi's single-number yields cannot attract institutional capital flows needed to build a real credit market without senior/mezzanine/equity pricing, and only 4 of 160+ teams are addressing this gap.

Markets are mispricing Coinbase as a crypto exchange when it should be valued as financial infrastructure for agentic commerce, with x402 integration into Amazon Bedrock providing enterprise validation.

Paying ransoms like Instructure did incentivizes future attacks and offers no absolute certainty that stolen data won't be leaked.

Lights-out AI codebases should be treated as a design target requiring new verification infrastructure, not an inevitability — because prompts aren't formal specs and AI-checks-AI pipelines are bolt-on rather than first-class CI.

The risks of not using AI outweigh the risks of using it, as evidenced by 79% of IT leaders exploring AI for incident management despite 74% citing security risks as a major barrier.

Despite widespread belief that JSON-LD schema boosts AI visibility, a study of 1,885 pages found it produced no meaningful citation gains and brands should not expect schema alone to drive AI citation improvements.

Frequent AI use reduces critical thinking and emotional depth, and brands that adopt it reactively prioritize speed over insight, risking cultural flattening as globally trained models default to familiar patterns.

The theme of the Musk vs. Altman trials is basically 'everyone vs. Google' because OpenAI was founded as a non-profit to stop Google from owning all of AI.

The end state of agentic intelligence should not be typing commands into Telegram forever — it should be the removal of every barrier between you and the technology, making ambient intelligence finally a reality.

Senior developers fail to communicate expertise because they frame problems in terms of complexity management rather than translating solutions into uncertainty reduction language the rest of the company understands.

Projects built as personal challenges without a solid use case will likely be abandoned as maintainers lose interest, warning builders not to lose sight of what made them successful.

Almost everything new in Android's AI overhaul will arrive via Play Services or app updates rather than as part of Android 17, suggesting Google is prioritizing speed of rollout over OS-level integration.

Brand design is not just vibes and aesthetics — HubSpot's rebrand drove measurable upswings across every marketing channel, proving design has real business impact.

Contradictory signals — mass layoffs citing AI alongside rising software job openings — make it genuinely unclear whether we're witnessing real disruption or another hype cycle

The engineer of the future is more likely to work at a pharma company than at Meta, as domain-specific expertise becomes the core differentiator

AI agents will multiply the number of workers using enterprise software rather than eliminate jobs, and the SaaSpocalypse narrative is wrong

Bitcoin is unusually tied to NASDAQ right now, making crypto highly sensitive to AI stock sentiment swings.

The current AI stock rally may be the final frothy phase before a reset, drawing parallels to the 1999 dot-com bubble.

With Pokémon cards appreciating ~140% over two years, the author calculates he will break even on his $1,000 Poliwrath purchase in 9.25 months.

The author warns that onchain gacha is dangerously addictive and explicitly asks readers to forward the newsletter to his wife to stop him from spending more.

Collector Crypt is a rare onchain success story that meaningfully drove the author's first onchain transaction in months, suggesting gacha mechanics are good for crypto adoption.

The NFT recovery question has shifted from 'is anything happening?' to 'how far will this go?' — the ground ahead seems firmer and more promising than it has in a while, with 30D breakouts, doubled wallet activity, and new verticals emerging.

U.S. inflation hit 3.8% in the latest reading — the highest since May 2023 — with oil cited as the primary culprit.

Diversification within a crypto-heavy portfolio is largely cosmetic because all crypto assets sell off together, making true diversification require assets outside the crypto correlation bucket.

Self-driving cars will not disintermediate Uber — AV fleets need demand aggregation during peak hours and complex routes, making Uber's network more valuable, not less, as robotaxis scale.

Amazon's internal AI token leaderboards created perverse incentives, with employees building pointless agents just to inflate usage metrics rather than drive genuine productivity.

Warsh historically prefers lower rates but is walking into a perfect storm of high oil and hot inflation, making his path uncertain.

Owning chip factories positions investors to win regardless of inflation because demand for compute is growing faster than inflation.

The author is leaning optimistic on the Trump-Xi meeting, reasoning that both leaders would cancel if they weren't confident it would yield tangible results.