Romane Robb
Romane Robb
@romanerobb.com
Romane Robb
ROBB
Entrepreneur · Investor · Creator
BUILDING IDEAS
INTO INSTITUTIONS.

Romane Orlando Robb

@romanerobb.com
Father. Husband. Serial entrepreneur behind 37+ ventures. Investor, software engineer, filmmaker and author. I talk business, private equity, venture capital, technology, media and the work of building.
37+ventures2,000+subscribers50+companies invested in
Romane Robb
@romanerobb.com · Oct 8
•••
I build companies, tell stories and invest in ideas with the power to last.

This is where I share the thinking behind the work—business, technology, media, family and everything I’m learning along the way.
Currently building
Robb Holding Company · GovConUnited · CrewUp · LeadsyCRM · Robb Entertainment · ameliazoe· AylaVictoria · Robb Investments · projekx · tikits.co · Robb Ventures · Robb Building Materials · AylaZoe Soap Company · and more!
Romane Robb
@romanerobb.com · Oct 29
•••
Anthropic’s $80.7B net loss is mostly paper, a non-cash charge that grew because the company got more valuable. Read the filing, not the graphic.
Bloomberg Tech graphic of Anthropic first-half results: revenue $16.3B vs $1.16B, operating expenses $17.0B vs $3.3B, operating loss, net loss -$80.7B. Source: company filings.

Let me be direct. The graphic making the rounds this week shows Anthropic’s first-half numbers, and the take spreading with it is that the AI boom is a mirage because the company behind Claude “lost” $80.7 billion in six months. I read the filing instead of the graphic. The filing tells a completely different story: a business that grew revenue 14 times in one year, cut its real operating burn down to almost nothing, and booked the biggest paper loss in tech history mostly because its own success made its promises to investors more expensive. That is not a collapse. That is what winning looks like under accounting rules.

Bloomberg Tech graphic of Anthropic first-half results: revenue $16.3B vs $1.16B, operating expenses $17.0B vs $3.3B, operating loss, net loss -$80.7B. Source: company filings.
The Bloomberg Tech graphic from Anthropic’s S-1 numbers. Source: Bloomberg Tech / company filings.

Read the Filing, Not the Graphic

The television graphic shows revenue of $16.3 billion for the first half of 2026, up from $1.16 billion in the first half of 2025. That part is real, and it is staggering. But look at the operating loss row. The graphic shows a $2.1 billion operating loss for 2026. Go to the actual S-1 and the operating loss narrowed to roughly $693 million. The $2.1 billion figure is the year-ago scale of red ink, not this year’s. In margin terms, Anthropic ran a GAAP operating margin of minus 4 percent in the first half of 2026, against minus 184 percent a year earlier. On the company’s adjusted measure, it actually made a $1.6 billion operating profit.

One row of a TV graphic versus the audited filing. Operators who read filings eat operators who read graphics. It has always been that way.

Where Did $80.7 Billion Go? Nowhere.

Here is the part the headline crowd skips. Roughly $72.4 billion of the $80.7 billion net loss is a non-cash charge tied to Anthropic’s convertible notes. The company raised money from investors using notes that convert into shares later. Accounting rules say those notes have to be marked to their fair value. As Anthropic’s valuation climbed toward the reported $2 trillion mark, the notes became more valuable on paper, and that increase gets booked as a loss. Read that again: the better the company performs, the bigger this “loss” gets. Not one dollar of that $72.4 billion left the building.

And the cash tells its own story. Anthropic held about $131.7 billion in cash and short-term investments on June 30. Companies that are bleeding out do not sit on $131.7 billion. The loss is an accounting shadow cast by a valuation explosion. Shadows are not wounds.

The Numbers That Actually Matter

Strip away the paper and this is what the filing actually shows:

  • Revenue: $16.29 billion in six months, against $1.16 billion in the same period a year earlier. Fourteen times growth, year over year.
  • Acceleration: $4.7 billion in the first quarter, then $11.6 billion in the second. The curve is bending up, not flattening out. Third quarter revenue reportedly hit $16.8 billion, more in one quarter than in the entire first half.
  • Context: the first half of 2026 alone brought in more than triple all of 2025, which totaled $4.59 billion.
  • Customers: more than 11,900 customers generated at least $100,000 each in the last year. More than 1,900 generated at least $1 million. About 250 generated more than $10 million. That is enterprise money at enterprise scale.
  • Efficiency: roughly 5,300 employees producing over $6.6 million of revenue per employee, up from $1.4 million a year ago. That one ratio is the AI era in a single number.

Paper Losses vs. Cash Losses

Every operator needs this distinction burned in. A cash loss is money out the door: compute, payroll, rent, the real cost of running the machine. A paper loss is an entry in a ledger. Anthropic’s real cash story is about $9 billion spent on computing and infrastructure in the first half, against $16.3 billion of revenue coming in, with the operating burn shrinking toward zero.

Then look at the commitments: roughly $500 billion in compute agreements spread over the next decade, with Broadcom, Google, Amazon, Microsoft, and AMD. Companies do not sign half a trillion dollars of infrastructure deals because they are dying. They sign them because demand is pulling product out of them faster than they can build capacity. Know which losses are real in your own books too. Depreciation, mark-to-market swings, and one-time charges can make a healthy business look sick and a sick business look healthy. The cash never lies. Follow the cash.

What This Means for Everyone Building on AI

I build with AI every day, and I have said it plainly: AI is now the primary way to build. This filing is the weather report for that world. The model layer is becoming an economy of its own, measured in hundreds of billions of infrastructure and tens of billions of revenue, and the company at the center of it runs at over $6.6 million of revenue per employee. That ratio changes what a small team can be. When the platform underneath you is that efficient, a five person company can punch like a five hundred person company, if the five people actually build instead of talking about building.

It also settles a question I hear constantly: is this real, or is it hype with good marketing? Hype does not produce 11,900 customers paying six figures a year. Hype does not triple a full year’s revenue in six months. Hype does not sign $500 billion in compute contracts. Revenue is the receipt. Anthropic just showed the receipt.

The Bottom Line

Anthropic filed to go public on Nasdaq under the ticker ANT, in what could be the largest IPO in history, with reports pointing to a raise of up to $100 billion at a valuation of up to $2 trillion. The headline number everyone will repeat is the $80.7 billion loss. The number that matters is 14: revenue multiplied by fourteen in a single year while the real burn shrank to a rounding error.

Anthropic did not lose $80.7 billion. It got more valuable so fast that the accounting turned its success into a loss on paper. Read the filing, not the graphic. Results over headlines, every single time.

Romane Robb
@romanerobb.com · Oct 29
•••
Small business owners pay monthly for software they could own outright. Here is why building your own tools with AI beats renting subscriptions forever.
Small business owner building his own custom business dashboard with an AI coding assistant while setting aside a stack of software subscription invoices

Here is a number most business owners never sit down and add up: the total they pay every month to rent software. Not own. Rent. Invoice tools, scheduling tools, inventory tools, customer follow-up tools, dashboards, forms, reports. None of it is theirs. The day they stop paying, it all disappears, and everything their business ran on turns out to have been a subscription with a login screen. That made sense when building your own software required a development team and a six figure budget. It does not make sense anymore. AI moved the ability to build directly into the owner’s hands, and owners who understand ownership in every other part of their business should understand it here too: build your own tools, own them outright, and stop paying rent on the systems your company runs on.

The Subscription Trap Nobody Adds Up

Software subscriptions are priced to feel small. Twenty dollars here, fifty dollars there, per user, per month. It feels harmless until you multiply it by a team, by twelve months, by five years. A tool that costs $60 per user per month costs a ten person business $36,000 over five years. Most owners run five to ten of these tools at once. That is real money leaving the business forever, and at the end of it the business owns exactly nothing.

Then the trap tightens. Prices go up, and they always go up. Features you depend on get moved behind a higher tier. The plan you built your process around gets renamed, repackaged, or discontinued. Your data sits in their system, in their format, and exporting it is just painful enough that you stay. That is not a vendor relationship. That is a landlord relationship, and the rent never stops.

  • Per user pricing punishes you for growing your own team
  • Price increases arrive on their schedule, not yours
  • Your business data lives in their format, under their control
  • Stop paying and the tool, and often the records inside it, are gone

You Are Building Equity in Someone Else’s Company

Every business owner understands the difference between renting a building and owning one. Rent is an expense that builds nothing. A mortgage builds equity in an asset you control. Software works the same way, and it amazes me how many sharp owners who would never rent forever in real estate happily rent forever in the systems their entire operation depends on.

When you build your own tool, every improvement you make is an asset. The custom dashboard that shows your numbers the way you actually think about them. The follow-up system that matches how your customers actually buy. The inventory view built around your real workflow instead of a generic one designed for everybody and therefore perfect for nobody. These things make your business faster, and they belong to you. If the business is ever sold, systems you own are part of what a buyer is paying for. Subscriptions you rent are worth zero on that day.

AI Moved the Build Button to the Owner’s Desk

The honest reason owners rented software for decades is that the alternative was worse. Building custom software meant hiring developers, managing a technical project you did not fully understand, and spending serious money before you saw anything work. Renting somebody else’s finished product was the rational move. I made that move myself, many times.

AI changed the math completely. The person who knows the business best, the owner, can now describe exactly what a tool needs to do and have it built: the fields, the workflow, the reports, the alerts. Not a generic product shaped for a million strangers. Your tool, shaped for your operation, built at the speed of your own thinking. The bottleneck is no longer technical skill or budget. The bottleneck is whether the owner can clearly say what the business needs, and that has always been the owner’s real job.

This is the part most people are still sleeping on. They compare the subscription price to the old cost of custom development and conclude renting wins. That comparison is dead. The new comparison is a monthly fee, forever, against building the thing once and owning it.

Build the Tools That Touch Your Money First

You do not have to rebuild everything at once. Start where fit matters most and the generic products serve you worst:

  • Scheduling and booking that matches how your operation actually runs
  • Customer follow-up systems that speak in your voice and run on your timeline
  • Inventory and order views built around your products, not a template
  • Dashboards and reports that show the numbers you check every morning, in one place
  • Internal trackers for jobs, batches, leads, and projects that currently live in spreadsheets and group chats

Every one of these is a tool where your process is the product. Generic software forces your process to fit its boxes. A tool you build fits your process exactly, because it was designed by the only person who truly knows it.

Rent the Vault, Own the Workshop

Let me be precise, because ownership is a strategy, not a religion. Some things should stay rented. Payment processing, banking connections, and anything carrying serious legal or security liability comes with compliance burdens that specialists exist to carry. Rent the vault.

But the workshop, the everyday tools your team lives in, the systems that hold your process and your data and your way of doing business, that should be yours. Own the workshop. The line is simple: if a tool’s main job is to express how your business works, build it. If its main job is to absorb regulated risk, rent it.

The Bottom Line

Business owners are builders by definition. You built the company, the team, the customer base, the brand. Renting the software that runs all of it was a compromise from an era when building was out of reach. That era is over. AI put building within reach of the owner, and the owners who move first will run leaner, move faster, and actually own the systems their growth depends on, while everyone else keeps paying monthly for the privilege of building someone else’s company.

Add up your subscriptions tonight. Look at the five year number. Then ask yourself what that money buys if you spend it building tools you own instead. I already know my answer. Results over rent, every time.

Romane Robb
@romanerobb.com · Oct 28
•••
Discover the benefits of AI web development vs a human developer: lower cost, faster delivery, 24/7 output, and consistent quality on every build today.
AI web development workspace with an AI interface assembling a website

Let me be direct: AI web development is the primary way to build websites and applications now, and the gap is not close. I say that as someone who has paid human developers and now builds with AI every single day. AI is cheaper. AI is more efficient. AI never stops working. It does not complain, it does not call in sick, and it does not bill you for a meeting about the meeting. When you compare the benefits of AI in web development against the old model of hiring a human developer, the result is the same on every line that matters: cost, speed, output, and consistency. Here is the full breakdown.

The Cost Difference Is Not Even Close

Hiring a human developer is one of the most expensive decisions a business can make. A single full time developer salary runs from $80,000 to well past $150,000 a year, and that is before benefits, payroll taxes, equipment, software licenses, recruiting fees, and the management time it takes to keep one person on track. Agencies charge $100 to $200 an hour. Freelancers charge less per hour and still bill for every revision, every call, and every delay.

AI web development flips that math completely. For the cost of a monthly subscription, you get a builder that writes code, designs pages, fixes bugs, and produces content on demand. There is no salary. There is no benefits package. There is no invoice padded with project management hours.

  • Human developer: five to six figures per year, every year
  • Agency: $100 to $200 per hour, billed in increments
  • AI web development: one flat subscription, output on demand
  • Revisions with a human: a new charge. Revisions with AI: included and instant

For a business owner, that difference is not a small optimization. It is the difference between building one website a year and building an entire portfolio of websites, stores, and applications in the same budget.

Speed and Availability: AI Never Stops Working

Human developers work business hours, in one time zone, with weekends off, holidays off, vacations, and sick days. A project that should take a week stretches into a month because the work only happens in small daily windows between meetings and days off.

AI works 24 hours a day, 7 days a week, 365 days a year. It never sleeps, never takes a break, and never stops working. When an idea hits at 3 in the morning, the build starts at 3 in the morning. When a page has to go live before a product launch, AI delivers in hours instead of sprint cycles.

  • No weekends, no holidays, no sick days, no vacation gaps
  • No time zone delays and no waiting for the next business day
  • First drafts in minutes, finished builds in hours
  • Multiple tasks move forward at the same time, around the clock

Speed is a competitive weapon. While a competitor waits on a developer’s calendar, your website is already live, already indexed by search engines, and already selling.

Efficiency and Resource Use

A human developer consumes far more than a salary. They need onboarding, documentation, hand holding, code reviews, and constant direction. Every misunderstanding becomes rework, and every round of rework becomes another billable cycle.

AI uses very few resources. Give it clear instructions and it executes. There is no office space, no hardware to buy, no training period, and no ramp up time on a new project. One person directing AI can produce the output of a small development team, because the AI handles the coding, the layout, the troubleshooting, and the repetitive work that eats a human team’s entire week.

The efficiency compounds over time. AI does not forget how your project is structured. It does not need to be re-briefed after a long weekend. It picks up exactly where the last session ended and keeps building, which means your momentum never resets.

Consistency and Instant Revisions

Human output depends on the human. Energy, mood, focus, and experience change from day to day, and the work changes with them. One page comes out polished, the next one comes out rushed. Deadlines slip because someone had a bad week, a bad night, or a bad attitude.

AI delivers the same quality bar on the first task and on the thousandth task. The standard does not move. And when something needs to change, the change happens right now. A new color scheme, a restructured page, a rewritten function, a completely different layout: revisions that take a human developer days of back and forth take AI a matter of minutes.

There is also a benefit nobody likes to say out loud: AI does not complain. No pushback, no ego, no attitude about feedback, and no drama. You ask, it delivers. That alone saves more time, money, and energy than most businesses ever calculate.

Scalability on Demand

Hiring scales slowly and painfully. Every new project means recruiting, interviewing, and onboarding, which takes months before any new output appears. Replacing an underperformer costs even more time and money.

AI scales instantly. One website or fifty websites, the capacity is there the moment you need it. You can build a store, a blog, a booking platform, and a company website in parallel, and launch all of them while a traditional team is still writing the project plan for the first one.

  • Launch multiple projects at once without hiring anyone
  • Scale output up during busy seasons and down when things are quiet
  • No recruiting, no onboarding, and no HR overhead at any size
  • Small businesses can ship with the output of a large company

Deep Knowledge on Tap

A human developer knows what one person can learn in one career: a few languages, a few frameworks, and whatever their past jobs happened to expose them to. Every new technology is a learning curve, and you are the one paying for that curve.

AI brings the combined knowledge of millions of projects to your build from the very first prompt. HTML, CSS, JavaScript, PHP, Python, databases, APIs, eCommerce, SEO, security best practices, and accessibility: it is all there, current, and ready to apply. AI app development and AI website building draw on patterns proven across the entire internet, not on one person’s resume.

There is no training cost and no knowledge gap. Ask for a feature and the AI already knows several ways to build it, and which one fits your stack best.

Frequently Asked Questions

Is AI web development cheaper than hiring a human developer?

Yes, by a wide margin. A human developer costs a full salary plus benefits, taxes, equipment, and overhead. AI web development runs on a flat subscription and produces output on demand, with revisions included at no extra charge.

Can AI build a complete website or app by itself?

Yes. An AI website builder can plan, code, design, and launch a complete website or application, including pages, features, databases, and integrations. You direct the vision, and the AI does the building.

How fast can AI deliver a website?

AI produces working drafts in minutes and finished websites in hours or a few days, depending on size. Because it works around the clock, there are no business hour limits slowing the build down.

Is AI the primary way to develop applications now?

Yes. Between the cost, the speed, the consistency, and the 24/7 availability, AI app development is the primary method for anyone who measures results instead of habits.

The Bottom Line

The AI vs human developer debate is over for anyone who has actually run the numbers. AI web development is cheaper, faster, more efficient, and more consistent. It never stops working, it never complains, and it uses a fraction of the resources a human hire requires. Human developers had a long run as the only option. They are not the only option anymore, and they are not the best one.

I build with AI because I care about output, and output is what grows a business. If you are still paying premium rates and waiting weeks for basic work, you are competing against people who are not. Make AI your primary way to build, and let the results speak for themselves.

Photos

Moments from business, media, travel, events and the work behind the scenes.

Podcast Episodes

Original conversations about business, music, media, ownership and the creative process.
The Business · Featured Episode

How Entrepreneurs Turn Ideas Into Institutions

Romane breaks down the systems, patience and operating discipline required to build companies designed to last.
Music Publishing Explained · New Episode

Who Really Owns a Song?

A practical look at master recordings, publishing rights, royalties and why ownership matters for independent artists.
Artist Process · Latest Episode

From Blank Page to Finished Record

A behind-the-scenes conversation about songwriting, production, collaboration and finishing creative work.
On Tour · Latest Episode

The Business Behind Live Music

How artists, managers, promoters and venues work together to take a show from routing to the stage.

Videos

Business insights, original series, interviews, field visits and behind-the-scenes stories.

Pocono Snake & Animal Farm 2022

Field visit · Romane Orlando Robb
BUILDING
IN PUBLIC

Building Companies Across Multiple Industries

Entrepreneurship · Business strategy
FOUNDER
NOTES

What Founders Should Know About Ownership

Private equity · Venture capital · Control
Search

Join the conversation

Get new essays, ideas and company updates in your inbox. Join 2,000+ subscribers.