Michael hochstat

data maker

[Michael hochstat @HochstatMichael]

Knowledge is free

https://t.co/X9QqbLjkgv

[Michael hochstat @HochstatMichael]

https://t.co/0YGyDJ5A04

[Michael hochstat @HochstatMichael]

@JakeBrownLive Jerry Cosbys and beef steak Charlie’s

[Michael hochstat @HochstatMichael]

AI isn’t killing crypto. But it is sucking a lot of the oxygen out of the room.

A few years ago, crypto was the dominant story in venture capital and tech investing. Today, AI has taken that crown.

The numbers tell the story:

• Nearly 40% of crypto VC funding now goes to companies that are also building AI products, up from just 18% a year earlier.
• AI is capturing the overwhelming majority of global venture investment.
• Crypto miners are increasingly repurposing infrastructure, GPUs, and power capacity for AI and high-performance computing.
• Capital that once chased speculative altcoins is now flowing into AI infrastructure, chips, data centers, and model development.

This doesn’t mean crypto is dead.

Bitcoin, Ethereum, stablecoins, tokenized real-world assets, and DeFi continue to attract billions in capital and users. Stablecoin transaction volumes are reaching record levels, and blockchain infrastructure keeps advancing.

But AI has become the new gravitational center of technology investing.

What’s fascinating is that the most promising crypto projects today aren’t competing with AI—they’re enabling it.

Decentralized GPU networks, distributed compute, AI agent economies, verifiable data markets, and DePIN projects are attracting attention because they sit at the intersection of both megatrends.

The lesson?

When a new technology wave arrives, capital follows the biggest opportunity. Right now, AI is that opportunity.

Crypto isn’t disappearing. It’s evolving—and increasingly becoming part of the AI story rather than a separate one.

The next trillion-dollar winners may not be AI companies or crypto companies.

They may be the companies building where the two worlds collide.

#AI #Crypto #Bitcoin #Blockchain #DePIN #VentureCapital #ArtificialIntelligence #Technology #Innovation #Web3

[Michael hochstat @HochstatMichael]

🍞 The Easiest Bread You’ll Ever Make: 3-Ingredient Beer Bread

No yeast. No kneading. No waiting for dough to rise.

If you can stir three ingredients together, you can make a fresh loaf of homemade beer bread in about an hour.

Ingredients
• 3 cups self-rising flour
• 12 oz beer (lager, amber, or your favorite brew)
• ½ cup sugar (or reduce to 3 Tbsp for a less sweet loaf)

Optional: 3 Tbsp melted butter for a golden, crispy crust.

Directions

Preheat oven to 375°F (190°C).
Grease a 9x5-inch loaf pan.
Mix flour and sugar in a bowl.
Pour in beer and stir until just combined. Don’t overmix.
Transfer batter to the loaf pan and top with melted butter if desired.
Bake for 45–55 minutes until golden brown and a toothpick comes out clean.
Cool for 10 minutes, slice, and enjoy.

Why it works:
The carbonation in the beer combines with the leavening in the self-rising flour to create a light, airy texture—without yeast or kneading.

Easy upgrades:
🧀 Cheddar cheese
🌿 Fresh herbs
🧄 Garlic and parmesan
🌶️ Jalapeños

Fresh bread doesn’t get much simpler than this. Perfect for soups, chili, BBQs, or just warm with butter straight from the oven.

#Baking #HomeCooking #BeerBread #EasyRecipes #Foodie #ComfortFood #BreadMaking #CookingAtHome

[Michael hochstat @HochstatMichael]

https://t.co/ZZNyBVBrLO

[Michael hochstat @HochstatMichael]

Goose is one of the most impressive open-source AI agents available today.

Developed by Block (Jack Dorsey’s company) and released in 2025, Goose runs locally on your device and works with a wide range of LLMs—including local models through Ollama as well as providers like OpenAI and Anthropic.

What makes it stand out?

• Autonomous project building: Give Goose a high-level prompt such as, “Build me a website like YouTube,” and it can create the project structure, write code, install dependencies, run commands, debug issues, and iterate toward a working solution.

• Free and open source: Released under the Apache 2.0 license, Goose itself is completely free to use.

• Local and private: It runs on your machine, keeping your code and data under your control.

• Highly extensible: Goose integrates with dozens of tools and services through open standards such as the Model Context Protocol (MCP), enabling connections to databases, browsers, GitHub, and more.

The reality is that AI agents like Goose won’t instantly build production-scale applications without oversight. Complex software still requires testing, refinement, and human judgment.

But the direction is clear: the barrier to creating software is dropping dramatically.

We’re moving from a world where building required years of coding experience and large teams to one where a clear vision, domain expertise, and effective AI collaboration can turn ideas into working products faster than ever before.

In this new environment, ideas, creativity, and execution may become even more valuable than technical implementation alone.

Have you experimented with Goose or other autonomous AI agents yet? What has your experience been?

[Michael hochstat @HochstatMichael]

What management actually said
Alphabet said Q1 2026 capex was $35.7B, with the “overwhelming majority” going to technical infrastructure:
“CapEx was $35.7 billion in the first quarter, with the overwhelming majority of this spent in technical infrastructure to support the AI opportunities we see across the company. Approximately 60% of our investment in technical infrastructure this quarter was in servers, and 40% was in data centers and networking equipment.”
— Anat Ashkenazi, GOOG Q1 2026
They then raised full-year capex guidance:
“We are updating our full year 2026 CapEx guidance range to $180 billion-$190 billion, up from our previous estimate of $175 billion-$185 billion…”
— Anat Ashkenazi, GOOG Q1 2026
And the kicker:
“We expect our 2027 CapEx to significantly increase compared to 2026.”
— Anat Ashkenazi, GOOG Q1 2026
My read
This is Alphabet explicitly telling the market that AI infrastructure is becoming the company’s new cost of admission. The important point is not just that 2026 capex is massive — it is that management is guiding to another significant step-up in 2027 before investors have seen a clean return profile on the 2026 spend.
The bullish interpretation: Alphabet is capacity-constrained, not demand-constrained. Sundar said cloud revenue would have been higher if Google had enough compute capacity, which supports the argument that capex is chasing real demand rather than speculative buildout.
The bearish interpretation: the historical Alphabet model — extremely high-margin search monetization with asset-light economics — is being partially replaced by an AI model that requires hyperscale physical infrastructure, higher depreciation, energy costs, and lower near-term free-cash-flow conversion.
The most important financial consequence: depreciation is now a structural https://t.co/srHx2LjLlq explicitly warned that technical infrastructure investment will pressure the P&L through higher depreciation and data-center operating costs such as energy. That means even if revenue accelerates, margin expansion may be harder than the market wants.
Bottom line
Alphabet is no longer just an advertising + cloud software compounder; it is becoming one of the world’s largest AI infrastructure builders.If the AI demand curve is real, this spend can be justified. If monetization lags, the market will eventually treat this as overbuilding.
The trade-off is stark: Alphabet is buying strategic relevance in AI with free cash flow.That is probably necessary — but it makes the stock’s future multiple more dependent on evidence of AI revenue conversion, not just AI usage or model leadership.
Used 2 sources from GOOG

[Michael hochstat @HochstatMichael]

🚨 AI Content Wars Are Escalating

A new lawsuit from CNN against Perplexity AI is becoming another major flashpoint in the battle over who owns value in the AI era.

According to the complaint, CNN alleges that Perplexity used more than 17,000 CNN articles, videos, and images without permission to help power AI-generated answers that compete directly with the original content.

Perplexity’s response: “You can’t copyright facts.”

While that argument may be legally relevant, the bigger business question is becoming impossible to ignore:

If AI companies can ingest, summarize, and monetize content created by others, who captures the economic value?

This isn’t just about media companies. It’s a strategic issue for every organization that produces proprietary knowledge, research, data, expertise, or intellectual property.

Business leaders should be asking:

✅ What knowledge assets do we truly own?

✅ How can they be protected or licensed?

✅ What competitive advantages can’t be easily scraped and reproduced by AI?

The CNN-Perplexity case joins a growing list of disputes involving publishers, content creators, and AI firms, highlighting a fundamental challenge of the AI economy: the tension between open access to information and the value of creating it.

The companies that thrive in the AI era may not be those with the most content—but those that best understand, protect, and monetize their unique knowledge assets.

#AI #GenerativeAI #ArtificialIntelligence #BusinessStrategy #DigitalTransformation #Media #Innovation #IntellectualProperty

[Michael hochstat @HochstatMichael]

🚨 The uncomfortable truth about AI in the legal industry: most projects still fail.

While firms are pouring billions into AI, the data suggests that 80%+ of legal AI initiatives fail to deliver the expected ROI.

Why?

Because AI adoption is rarely a technology problem. It’s an execution problem.

📊 Recent industry findings show:

• More than 80% of legal professionals expect AI to have a high or transformative impact within the next five years.
• Yet only a small minority of firms have a clearly defined AI strategy.
• Many organizations remain stuck in the “pilot phase” without reaching production-scale deployment.
• Adoption, training, workflow integration, and ROI measurement continue to be the biggest obstacles.

The biggest reasons AI projects fail:

❌ Poor user adoption and change management
❌ Resistance from established workflows and leadership teams
❌ Weak integration into day-to-day legal processes
❌ Difficulty measuring business value and ROI
❌ Data quality, privacy, and governance challenges
❌ AI initiatives launched because of hype rather than clear business objectives

This is why many industry observers are calling 2026 the “groundwork year” for legal AI.

The focus is shifting from experimentation to building the foundations that actually generate long-term value:

✅ AI strategy
✅ Data infrastructure
✅ Lawyer training
✅ Workflow integration
✅ Governance and risk controls
✅ Clear success metrics

The firms that win won’t necessarily have the best models.

They’ll have the best implementation.

That’s what makes Kirkland & Ellis’ $500 million AI investment so notable. The real story isn’t the size of the check—it’s the emphasis on building AI directly into legal workflows with extensive lawyer involvement.

AI is moving from a technology experiment to a core business capability.

The next phase of the AI race won’t be won by who adopts AI first.

It will be won by who operationalizes it best.

#AI #LegalTech #GenerativeAI #ArtificialIntelligence #LawFirmInnovation #DigitalTransformation #FutureOfWork #BigLaw #Leadership #TechnologyStrategy

[Michael hochstat @HochstatMichael]

@ImtiazMadmood My neighbor had one do that

[Michael hochstat @HochstatMichael]

https://t.co/uvfdXUd2bW

[Michael hochstat @HochstatMichael]

This sounds like science fiction, but it’s real.

GuRu Wireless, a Pasadena-based company spun out of Caltech research, has demonstrated wireless power beaming capable of keeping a drone airborne for 96 straight hours.

That’s 4 continuous days of flight.

The setup uses a 24 GHz millimeter-wave phased array system that beams RF energy from a ground transmitter directly to a receiver mounted on the drone.

In demonstrations:
• The drone operated untethered from roughly 30 feet away
• Power was dynamically adjusted while the drone remained in flight
• The system continuously recharged the onboard battery mid-air, dramatically extending endurance beyond traditional battery limits

For comparison:
Most commercial drones today only fly for about 30–45 minutes before needing to land and recharge.

This changes the equation completely.

The implications are enormous:
• Persistent surveillance and ISR missions
• Long-duration infrastructure inspection
• Autonomous security monitoring
• Continuous communications relays
• Eventually, potentially persistent delivery fleets or remote-area operations

And this is bigger than drones.

Wireless power beaming — whether through RF, microwave, or laser systems — is rapidly moving from experimental research into real-world deployment.

The bottleneck for many autonomous systems has always been energy.

If you remove the need to constantly land, recharge, or swap batteries, entirely new categories of robotics and autonomous infrastructure become possible.

Of course, there are still major challenges:
• Efficiency losses over longer distances
• Weather and line-of-sight limitations
• RF safety and regulatory approval
• Cost and scalability outside specialized use cases

Right now, this technology is mainly aimed at defense and high-value industrial applications.

But it’s another reminder that some technologies we assumed were decades away are arriving much faster than expected.

Power beaming is no longer just a sci-fi concept.

It’s becoming an engineering reality.

[Michael hochstat @HochstatMichael]

The sudden decline of “tokenmaxxing” may end up being one of the most important shifts in the AI industry.

For the past two years, many companies treated massive AI token consumption as a proxy for innovation and productivity. More prompts. More code generation. More usage. More spending.

Now the conversation is changing:
Enterprises are increasingly asking about efficiency, margins, ROI, and whether all this AI spending actually produces durable business value.

That shift matters.

A lot of the hottest AI companies are still growing explosively, but the economics underneath are getting harder to ignore.

• OpenAI continues to grow at extraordinary scale, but with enormous compute costs and ongoing losses tied to infrastructure expansion.
• Anthropic is seeing hypergrowth and improving margins, especially through enterprise coding adoption, but long-term profitability still depends on sustaining premium demand while inference costs fall.
• Google is rapidly closing gaps through Gemini and ecosystem integration.
• Chinese players like DeepSeek, Alibaba/Qwen, and ByteDance are accelerating commoditization with cheaper, increasingly capable models.

This is the key trend:
Foundation models are slowly moving from scarcity economics toward commodity economics.

Over time:
• Token prices likely collapse
• Margins compress
• Open-source models improve rapidly
• Enterprise buyers become more price-sensitive
• Value shifts upward into applications, agents, proprietary data, workflows, and distribution

That doesn’t mean the current leaders suddenly collapse.

It means the market is transitioning from:
“Who has the biggest model?”
to
“Who can build sustainable businesses on top of these models?”

The same pressure applies across the stack.

AI startups face rising infrastructure bills.
Hyperscalers are deploying hundreds of billions into capex.
Even Nvidia — despite still dominating — eventually faces normalization once AI hardware becomes less supply constrained and alternatives mature.

The winners of the next phase probably won’t just be the companies generating the most tokens.

They’ll be the companies that:
• deliver measurable productivity gains,
• build sticky ecosystems,
• own distribution,
• control infrastructure efficiently,
• and create products people repeatedly pay for.

The gold rush phase of AI is evolving into an efficiency and execution phase.

That’s a very different market.

[Michael hochstat @HochstatMichael]

🚨 Webflow Layoffs Signal the Next Phase of the AI Restructuring Era

Webflow, the San Francisco-based no-code website builder, reportedly conducted a major round of layoffs on May 28, 2026 — with employees abruptly locked out of company systems before receiving termination notices via email.

According to reports:
• CEO Linda Tong described the cuts as a necessary “restructure” at an “inflection point,” as Webflow shifts toward smaller, AI-focused teams.
• Employees reportedly lost access to Slack and internal systems early in the morning, with notifications arriving minutes later — a process the company framed as standard security protocol.
• One engineer reportedly called the situation a “bloodbath,” arguing leadership may be overestimating how much human work AI can realistically replace.
• This follows an earlier 2024 reduction of roughly 8% of staff.

But this story is bigger than Webflow.

The tech industry is now deep into a broad AI-driven efficiency wave. More than 140,000 tech jobs have reportedly been eliminated in 2026 already, with companies including Meta, Intuit, Cloudflare, Wix, and others restructuring around automation, profitability, and AI-first operating models.

What’s driving it?
• AI tools are rapidly automating workflows across design, coding, marketing, testing, and support.
• Companies are under pressure to improve margins after years of aggressive hiring and cheap capital.
• Leadership teams increasingly believe smaller organizations augmented by AI can move faster and operate leaner.

The challenge is execution.

System lockouts, minimal communication, and abrupt notifications may protect security and IP, but they also damage trust and morale. Across the industry, employees are raising the same concern: companies may be underestimating the gap between AI-assisted productivity and fully replacing experienced human teams.

This wasn’t a scandal. It was a strategic decision.

But it highlights the defining tension of the current AI transition:
Companies are prioritizing speed, efficiency, and adaptability — while workers absorb the uncertainty, disruption, and human cost in real time.

[Michael hochstat @HochstatMichael]

🚨 BREAKING: Meta is testing paid premium subscriptions across Instagram, Facebook, and WhatsApp.

Meta confirmed optional premium tiers are now being tested globally, adding advanced AI, privacy, and creator features while keeping the core apps free.

What’s reportedly included in early tests:

• Instagram:

Anonymous Story viewing
“Who unfollowed you” insights
Expanded audience controls and lists

• AI upgrades:

Higher usage limits for Meta’s generative AI tools
Access to advanced video remixing and creation features
Deeper integration with Meta’s AI agents and automation tools

• WhatsApp & Facebook:

Advanced messaging/business tools
Customization features
Enhanced privacy and control settings

Meta is effectively building a new recurring revenue engine beyond advertising.

The bigger story:
This is less about social media subscriptions — and more about monetizing AI infrastructure.

As AI compute costs explode, platforms are shifting toward:
• Subscription revenue
• Tiered AI access
• Premium productivity ecosystems

Meta appears to be following the path pioneered by:
• Snapchat+
• X Premium
• YouTube Premium
• Discord Nitro

But at a much larger scale given Meta’s billions of users.

Wall Street will likely view this as:
✅ Higher-margin recurring revenue
✅ Better AI monetization
✅ Reduced dependence on ad markets

Users?
Mixed reactions so far:
• Investors are excited
• Consumers are worried about “subscription fatigue”

One thing is becoming clear:
The era of completely free consumer internet platforms may be ending.

[Michael hochstat @HochstatMichael]

@TheMayorMatt Street meat

[Michael hochstat @HochstatMichael]

@Emily_Sundberg Mcsorley’s it’s why

[Michael hochstat @HochstatMichael]

🚨 BREAKING: NVIDIA just revealed some of its biggest strategic AI ecosystem bets yet in its latest Q1 2026 13F filing.

This wasn’t a normal investment portfolio update.

It was effectively a roadmap for where Jensen Huang believes the AI infrastructure stack is heading next.

Key moves from the filing:

• NVIDIA disclosed a massive new position in Coherent Corp.:
→ 7.79 million shares
→ valued at roughly $1.86B at quarter end

This follows NVIDIA’s previously announced ~$2B strategic investment and partnership focused on:

optical interconnects,
photonics,
advanced AI networking,
and U.S. manufacturing expansion.

Why this matters:

The next AI bottleneck is no longer just GPUs.

It’s:

bandwidth,
power efficiency,
rack-scale networking,
optical data movement,
and interconnect latency.

Photonics is becoming foundational to scaling frontier AI systems.

Meanwhile, NVIDIA also nearly doubled its stake in CoreWeave:

→ 47.2 million shares
→ worth roughly $3.66B
→ now one of NVIDIA’s largest disclosed holdings

That position reflects NVIDIA’s growing strategy of directly backing AI-native cloud providers that massively consume NVIDIA infrastructure.

CoreWeave has effectively become one of the most important “GPU hyperscalers” in the world.

The bigger picture is what stands out.

NVIDIA’s disclosed holdings increasingly resemble a vertically integrated AI ecosystem strategy:

• GPUs → NVIDIA
• AI cloud capacity → CoreWeave
• Optical networking + photonics → Coherent
• EDA/software stack → Synopsys
• Communications infrastructure → Nokia

This is not passive investing.

It’s ecosystem engineering.

NVIDIA appears to be using its balance sheet to accelerate:

AI infrastructure buildout,
supply-chain resilience,
optical networking adoption,
and demand creation for future generations of accelerated computing.

The market increasingly treats NVIDIA as a chip company.

But these filings suggest NVIDIA is positioning itself more like:
an AI infrastructure platform orchestrator.

That distinction may become extremely important over the next 3–5 years.

[Michael hochstat @HochstatMichael]

@RogerClark41 @eastendbar

[Michael hochstat @HochstatMichael]

@TommyBeer