Tomasz Tunguz

Student of Startups
Backer of 9 unicorns
Author of http://tomtunguz.com 
Subscribe https://ttunguz.kit.com/ef0a779b9c

Founder

I burned 84 million tokens on February 28th. Researching companies, drafting memos, running agents.

That’s running Kimi K2.5, a serverless model via API. At Claude or OpenAI rates — roughly $9 per million tokens blended — equivalent usage would cost $756 for a single day’s work. My peak days hit 80 million tokens. My average days run 20 million. Cloud inference at frontier-model pricing adds up fast.

This week, Alibaba released Qwen3.5-9B, an open-source model that matches Claude Opus 4.1 from December 2025. It runs locally on 12GB of RAM. Three months ago, this capability required a data center. Now it requires a power outlet.

A $5,000 laptop — a MacBook Pro with enough memory to run Qwen locally — pays for itself after 556 million tokens. At my usage rate, that’s about a month. At 20 million tokens per day, it’s four weeks.

After payback, the marginal cost drops to electricity.

It isn’t an intelligence compromise. Reasoning, coding, agentic workflows, document processing, instruction following : the 9B model matches December’s frontier across the board.

What changes when frontier intelligence runs locally? Everything I send to cloud APIs today — drafting emails, researching companies, writing code, analyzing documents — stays on my machine. No API logs. No third-party retention. No outages. No rate limits.

The tradeoff is parallelization. Cloud APIs handle thousands of concurrent requests. A laptop runs one inference at a time. For simple tasks — summarization, drafting, Q&A — that’s fine.

Queue them up. Let them run overnight. For complex agentic workflows that spawn dozens of parallel threads, local inference may not be worth the wait. The economics favor depth over breadth : fewer tasks, run longer, run cheaper.

Three months from data center to laptop. The buy-vs-rent math just changed.

Last quarter, my AI inference costs hit $100,000 annualized.

I started small. Six months earlier, I was spending $200 a month on Claude. Then I added three agent subscriptions : Codex, Gemini, & Claude Code. I was paying $600 a month.

Next I started using AI to transform my todo list into my done list, increasing tasks to 31 per day. $92 daily inference invoices started arriving. Then $400 per month on browser agents.

Within two quarters, my inference spend grew from $7,200 to $43,000 to over $100,000 run rate.

So I migrated to an open source model. It took a weekend. The key was building the right testing loops : I had six months of historical task data, so I could replay requests through the new model & hill-climb to parity with AI agents working through the night. By Sunday evening, they performed identically. At 12% of the cost.

I’m not the only one paying attention to this cost.

Technology companies are adding a fourth component to engineering compensation : salary, bonus, options, & inference costs. Levels.fyi pegs the 75th percentile software engineer salary at $375k. Add $100k in inference & the fully loaded cost is $475k. That’s 21% in tokens.

The question CFOs will pose : what am I getting for all this inference spend? Can I do it cheaper?

If the metric for a new cloud is gross profit per GPU hour, the employee equivalent is : productive work per dollar of inference.

For me, the answer is 31 tasks a day at $12k annually. The engineer still burning $100k? They’d better be 8x more productive!

Will you be paid in tokens? In 2026, you likely will start to be.

Databricks started later. It built a more complex architecture. It focused on unstructured data; images, documents, logs, audio. Though vast within the enterprise, this data had historically produced little insight. Too hard to process. Too messy to query. Too expensive to store in formats that mattered.

Snowflake took the opposite bet. Structured data. Clean tables. SQL queries that ran fast & returned answers executives could read. The market agreed. Snowflake went public at a $70 billion valuation. Databricks raised private rounds at half that.

Then AI arrived. Suddenly the data that was too messy to query became the data that models needed to train. Unstructured data wasn’t a liability. It was the asset.

Databricks started later. It built a more complex architecture. It focused on unstructured data; images, documents, logs, audio. Though vast within th

Open article: Databricks Overtakes Snowflake by Tomasz Tunguz • 2 min read

Last quarter, my AI inference costs hit $100,000 annualized.

I started small. Six months earlier, I was spending $200 a month on Claude. Then I added three agent subscriptions : Codex, Gemini, & Claude Code. I was paying $600 a month.

Next I started using AI to transform my todo list into my done list, increasing tasks to 31 per day. $92 daily inference invoices started arriving. Then $400 per month on browser agents.

Within two quarters, my inference spend grew from $7,200 to $43,000 to over $100,000 run rate.

Last quarter, my AI inference costs hit $100,000 annualized. I started small. Six months earlier, I was spending $200 a month on Claude. Then I added 

Open article: Will I Be Paid in Tokens? by Tomasz Tunguz • 2 min read

I burned 84 million tokens on February 28th. Researching companies, drafting memos, running agents.

That’s running Kimi K2.5, a serverless model via API. At Claude or OpenAI rates — roughly $9 per million tokens blended — equivalent usage would cost $756 for a single day’s work. My peak days hit 80 million tokens. My average days run 20 million. Cloud inference at frontier-model pricing adds up fast.

I burned 84 million tokens on February 28th. Researching companies, drafting memos, running agents. That’s running Kimi K2.5, a serverless model via A

Open article: Data Center Intelligence at the Price of a Laptop by Tomasz Tunguz • 2 min read

Data Center Intelligence at the Price of a Laptop

For the first time in venture history, three distinct channels share the liquidity burden roughly equally.

A decade ago, secondaries barely registered. They accounted for roughly 3% of exit value in 2015. Today they claim 31% : nearly $95b in the trailing twelve months.

The shift accelerated after 2021’s IPO bonanza. When public markets closed their doors in 2022, investors found alternative routes. Secondaries absorbed demand that would have flowed to traditional exits. When Goldman Sachs acquired Industry Ventures, the transaction signaled secondaries have arrived. Morgan Stanley followed with EquityZen, then Charles Schwab announced its acquisition of Forge Global. Wall Street recognized the structural change before most of venture did.

This matters for founders & investors. When IPOs dominated exits, fund models assumed a small number of public offerings would generate the bulk of returns.

Now liquidity arrives through multiple doors. A founder might sell secondary shares to patient capital while the company remains private. A GP might move positions through continuation vehicles. An LP might trade fund stakes on an increasingly liquid secondary market.

The 830 unicorns holding $3.9t in aggregate post-money valuation cannot all exit through IPOs. The math doesn’t work. At 2025’s pace of 48 VC-backed IPOs, clearing the unicorn backlog would take seventeen years. Secondaries provide a release valve that traditional exits cannot.

Companies like OpenAI have embraced this reality, running employee tender offers while voiding unauthorized secondary transfers. The largest private companies now manage their own liquidity programs rather than waiting for public markets.

Today, secondary liquidity concentrates in the top 20 names. SpaceX, Stripe, OpenAI. For the founder of company #50, the secondary market remains largely theoretical. For secondaries to succeed as a broad asset class, buyers must underwrite positions in companies without household recognition. As the market grows, this coverage gap becomes opportunity.

For LPs starved of distributions since 2022, the expansion of secondary channels offers hope. The $169b in cumulative negative net cash flows needs somewhere to go. More exit paths mean more opportunities to return capital.

When a Series B employee asks about liquidity today, the answer isn’t “wait for the IPO.” It’s “we’re planning a tender offer next year.”

A decade ago, secondaries were a footnote. Now they’re infrastructure. Liquidity flows where it can, not where tradition suggests it should.

Founder, Artisanal Ventures & Artisanal Talent Group, keeper of a genius pit bull. In service to tech Founders, family and friends.

With 33 years of experience in the software industry, I co-Founded two companies that complement each other: Artisanal Ventures, a venture firm that invests in and supports high performing clients of the other company, Artisanal Talent, an exec search firm that specializes in SaaS/AI/Security, etc.  My mission is to connect capital and great company building leaders with talented Founders and help them build the most iconic SaaS/Cloud companies in the history of software.

I spend all my time helping Founders, and assist my Partners heavily in an advisory capacity helping them close the best people in the industry, assist with fundraising, play broad advisory roles on all things people related, etc.

As Founder of Artisanal Ventures, I leverage a gigantic, global network of company builders, operators, and investors to source, diligence, invest in, and advise some of the best SaaS/Cloud companies in the market.  I have helped build multiple winners such as Abnormal AI, Anthropic, Snowflake, Databricks, DocuSign, MuleSoft, Confluent, Wiz, Cyera and many others and have an active portfolio of more than 40 companies in DevOps, Security, ML/AI, big data and SaaS.

As a Founder of Artisanal Talent, I built an A+ team of specialized recruiters who hire engineering, product, GTM, CEO/Board, and finance/HR leaders for our clients. We are the ONLY executive search firm whose core mission is total alignment with Founders and their backers. Our goal is not to extract as much fee revenue for searches as we can, like larger firms, and take on too many searches, desperate to generate more fee revenue.  Our goal is to take on less work and go way more deeply into our clients than “average”recruiters, with the goal being becoming an extension of Founders, and earning the right to invest our capital into their companies and help drive outcomes north of 10B.  Because our clients have generated a TRILLION dollars of value growth since we formed the company, it has an A+++ brand.   Guilt by association with mega winners!

We are also the single most selective search firm I’ve ever witnessed, by design.  Unapologetically picky about our clientele.  

Hobbies: time w my wife, kids, dogs and friends, traveling, philanthropy, guitar, and adrenaline sports like surfing, skiing, mtn biking, climbing, spearfishing, etc.  Grew up on a farm and shoveled my share of chicken droppings, raised by a Marine Corps fighter pilot, yada yada.  Drive giant trucks to support adrenaline habits.  

I don’t do career chats.  Don’t ask please.  I’m too old and three decades of that is enough.  

Andy Price

Get ready for a tidal wave of turnover.  We can start w easy targets called public companies cause their stocks got crushed in 25 and are going nowhere.  Why do you stay at any public company?  Some would argue they’re priced like startups and that’s a legit argument so plenty of upside rebuilding some of those.  But near term?  They’re gonna lose a gigantic amount of talent because they see no wealth creation ahead of them.  Some extremely high profile departures to hot startups are all over the industry. And it’s the tip of the iceberg.  

Toggle to startups. Plenty of turnover there.  Mostly due to internal dysfunction or you’re pre-AI, didn’t adapt, etc. But here is the biggest problem and it happens every Feb, like clockwork.  Founders/CEO’s get w their CRO’s and say “thanks for growing 100 pct last year despite tariff insanity and other friction.  Yay great SKO.  All yay’s.  Rainbows and butterflies abound.  Now let’s grow 150 pct.”  CRO says “I can’t do that due to gravitational reality called capacity.  Oh and the market won’t let us grow that fast.”  Or, “we need a second product and you have not delivered one so WTF am I supposed to do to hit your arbitrary number?”  This creates an unstable relationship between the CRO and the leader of the company, who is likely dealing with a bunch of delusional VC’s who over paid in the last round and are pressuring the Founder / CEO to defy gravity and rip. Don’t listen to your CRO?  They’ll quit.  And that’s disruptive as hell!

The correct approach is to have three players at the table: a CFO who has a clue and can help forecast. CRO who is financially astute: CEO who is objective and unemotional who hasn’t made promises to investors that they can’t keep.  If that group coverges on a number that’s exciting and yet plausible?  March ahead.  If you tell your CRO “go take out ISIS with a pellet gun!”  Be assured you’re gonna lose them AND miss the number.  Ouch.  All hell breaks loose.  Avoid that!

So don’t be delusional this month or whenever you forecast.  Be unemotional about this process and fair and yet aggressive.   That formula might lead to everyone being happy at the end of the year vs an internal civil war after the number was missed because it wasn’t forecasted properly in the first place.   Oh, and Founders raising money at stupid prices?  I really wish you studied the industry the way I do: sitting slightly above it looking down on the field and watching company after company from last gen get absolutely SMASHED by having raised at stupid prices and boxing themselves into an illiquid equivalent of pergatory.  Just stop already and think about the next round or the last round or the big picture or the fact that most IPO’s since 2018 are now trading in single digit multiple land.   And you’re at 200x.  Does not translate to reality.  And reality is  where you wanna live at all times, no exceptions.   

Good luck to all in what promises to be another wild year.

Founder of Artisanal Ventures & Artisanal Talent Group. Customer service rep for A+ entrepreneurs.

The latest insanity I’m seeing in the market?  Cloud / SaaS people thinking old comp models fit AI startups.  They don’t.  So get over it!  You’re not gonna get 3 pct of a company growing 300-500 pct at scale.  Probably not even 1 pct.  New age=new rules.   And yeah public market comps don’t correlate to anything in private markets but how many Anthropic’s are in public markets?  ZERO.  

There is no apples to apples comparison to last gen CO’s vs new ones.  Yes some AI startups have lousy margins and are full of it or dysfunctional or whatever.  But there are gigantic winners being built right now and insisting on percentages of something that will be more valuable than any company many of us have ever been associated with warrants a re calibration of our market values, financially speaking.   

We all have a choice.  Be in AI right now or be irrelevant.  If someone wants to be irrelevant over their ego needing 1.5 pct of a company that’s got 10-20x more upside than anything they’ve ever done, well, ok.  I’m sure ServiceNow or Salesforce will take you.  

Be smart about playing this next move.  Or spend the rest of your career regretting not participating in something truly transformational.  

I’ll close w this.  AI Founders look down on people who think they’re celeb’s.  They’re not impressed.   So they’ll happily take the next gen talent who’s willing to work 996, which, yeah, sucks, but a) won’t last forever and b) is one reason many of these players are moving so fast, etc.  The amount of money the early stage leaders (player / coaches) are making is completely LOCO.  So final advice is look at early growth too if you want crazy upside.  

Old rules are gone.  Just forget they ever existed.

No one loves European Founders more than I do, but the folks who give them advice?  They often need advice themselves.  Here's what we see:
--Use C- recruiters because they're cheaper.  And wonder why they get questionable talent before calling a tier one search firm.
--Lowball A+ exec's.
--Stay in Europe vs come to the US and hang w your key people, bond w them, etc.  This is a two way street.  If you're American and don't visit HQ in, say, Sweden?  Big miss.  IRL has been back for a while.  Get on that plane!

The winning formula to me is to combine best practices of both societies.  Now here's what I absolutely love about them.  Europeans show a lot less bravado, in my experience.  I have found most of them coachable, empathetic, elegant, adaptable.  YAY.  Fun to deal w folks like this!

Gonna shout out one of my favorites in Parloa. Malte Kosub is INCREDIBLE.  An amazing recruiter, super fun and energetic to talk to, no BS, an epic product in a crowded but gigantic market, listens but also knows what he wants and knows who is in charge without being condescending.  Classy, classy person.  Terrific formula.  Watch that guy closely.  Another is Max Junestrand at Legora.  Humble, wicked smart, hires really well, no drama, no turnover.  So many examples and to me, the common thread is that they find a way to blend the best of American culture and European culture, are pragmatic about bridging the different "norms," etc.  Could name so many more and will over time.  These two just popped into my ADHD mind as I was writing this.  But yeah.  As a native NorCal guy, I can say w confidence I'm super proud of our dominance in general, in AI, but there are PLENTY of A+++ companies getting formed all over the place and Europe is producing more than its share of them.

GO EUROPE!

taplio

The Battle

Tomasz Tunguz

Andy Price

Followers

Posts

Engagements

Likes

Performance ScoreTomasz Tunguz's LinkedIn Performance

Performance ScoreAndy Price's LinkedIn Performance

Ready to grow your LinkedIn presence?

taplio

The Battle

Tomasz Tunguz

Andy Price

Followers

Posts

Engagements

Likes

Performance Score.css-goub6s{font-size:var(--chakra-fontSizes-md);font-weight:500;color:var(--chakra-colors-gray-600);display:block;margin-top:var(--chakra-space-1);}Tomasz Tunguz's LinkedIn Performance

Performance ScoreAndy Price's LinkedIn Performance

Ready to grow your LinkedIn presence?

Performance ScoreTomasz Tunguz's LinkedIn Performance