The Kahnweiler Technique
David Peterson

Picasso’s 1912 deal with his dealer Daniel-Henry Kahnweiler reads almost like a term sheet.

The terms were as follows: Picasso offered exclusive rights to his work for three years at fixed prices. Kahnweiler agreed to buy everything Picasso produced. Picasso could choose five pieces per year that he’d keep for himself, but Kahnweiler had to buy everything else.

You may not have heard of him, but Kahnweiler was one of the most influential art dealers of the twentieth century. In June 2010, researchers at the Horizon Research Group investigated his approach to try to understand what made him so good. They landed on this early-stage, high-volume, long-hold strategy, captured so perfectly by the Picasso deal, which they named the Kahnweiler Technique.

What I found particularly interesting about the deal is that Kahnweiler was explicitly not trying to pick the one masterpiece among many duds. He knew he couldn’t predict what was going to take off, so he structured the deal to get as many shots on goal as possible.

And before you claim that Kahnweiler clearly didn’t have “taste” and should have made some big bets, it’s worth noting that Picasso himself couldn’t predict which of his pieces were going to be hits either (or else he would have kept only those for himself, as the contract stipulated!).

That’s the part I keep thinking about. 

All of us in the startup world talk about our vision for the future as if it is both easily predictable and inevitable. Scroll through X for even a few minutes and you’ll see breathless predictions about how exactly AI will rewrite everything.

But if even a creator like Picasso can’t tell what will work, how can the rest of us be so certain?

I was catching up with an old mentor a few weeks ago. He’s been a seed investor for almost twenty years. And, as he was reflecting on how his portfolio has evolved over time, he noted how unpredictable it all had been. The companies he thought would make his career had disappeared. The companies that had returned more capital than any others had been invisible to him at the time. The only thing that mattered (for both him as a seed investor and those founders he backed that eventually won out) was staying in the game long enough for wins to compound and the market to catch up.

It’s clear that we could all learn a thing or two from Kahnweiler. Not because he could see the future, but because he knew he couldn’t. Certainty may play well on X, but reality rewards people who keep their heads down and keep taking swings.

The AI-Native Enterprise Playbook
Ten real-time observations on a rapidly evolving playing field

No More Painting by Numbers
It’s the end of the “SaaS playbook.

The Age of Artisanal Software May Finally be Over
Every wave of technological innovation has been catalyzed by the cost of something expensive trending to zero. Now that’s happening to software.

Founders as Experiment Designers
David on why founders should run everything as an experiment.

ENTERPRISE/TECH NEWS

The Adolescence of Technology - Dario Amodei, founder of Anthropic, published his companion piece to 'Machines of Loving Grace' entitled The Adolescence of Technology: an essay on the risks posed by powerful AI to national security, economies and democracy—and how we can defend against them. He reassesses the timeline to superhuman systems at just 1-2 years and warns of the risks posed by what he calls a "country of geniuses in a datacenter," which could arrive as early as 2027. "We are considerably closer to real danger in 2026 than we were in 2023." Amodei references Claude's Constitution, which Anthropic published last week, and how they are thinking about 'growing' rather than building these systems to address the existential risks he outlines.

Flash Google - AJ Green’s AG+ newsletter calls out the recent ‘fundamental shift’ in cost-to-intelligence ratio, including Gemini 3 Flash. "The era of “bigger is better” is being challenged by “smarter and faster.” Having seen the internal telemetry, it’s clear Google has optimized Flash to solve the most painful bottleneck in AI today: the latency-to-cost ratio. When a $0.50 model can beat flagship performance on agentic coding, the market for expensive, slow reasoning models starts to evaporate. Google is essentially nuking the pricing floor for frontier intelligence, forcing OpenAI into a defensive posture."

Narrative Violation - SemiAnalysis published ‘From Tokens to Burgers: A Water Footprint Face-Off” with their calculations on water usage in data centres, outlining that Colossus 2, one of the world's largest AI datacenters, will use as much water/year as 2.5 average In-N-Outs. “Datacenter water usage is under increasing scrutiny, with projects even paused or canceled. See recent discussions in Arizona. We think the debate is overstated, as the numbers are often misleading and key variables are frequently ignored. By this we mean the cooling architecture (one of the main drivers of water usage), power source, location and local water scarcity or water source, among others. In addition, datacenter water consumption is often taken at face value and rarely put in perspective versus other everyday industries.” 

HOW TO STARTUP

Tanking Machines - The Wall Street Journal wrote about the ‘messy human drama’ behind Thinking Machines’ loss of three senior team members, who returned to OpenAI. The New York Times called it a ‘soap opera’. Founder Mira Murati previously left OpenAI in 2024 along with multiple colleagues including Barret Zoph, who was at the centre of last week’s drama. For fans of The Office, Chris Bakke tweeted that Thinking Machines (which recently raised $2bn at a $12bn valuation) is ‘The Michael Scott Paper Company of AI’. 

HOW TO VENTURE

State of Venture - Carta published their ‘State of Startups 2025’. Peter Walker, Carta’s Head of Insights posted the key findings including “𝗔𝗜 𝗮𝘁𝗲 𝟰𝟰% 𝗼𝗳 𝗶𝗻𝘃𝗲𝘀𝘁𝗲𝗱 𝗰𝗮𝗽𝗶𝘁𝗮𝗹 (𝘀𝗹𝗶𝗱𝗲 𝟭𝟬) and that share climbs to 61% if you just look at software startups. If you are building a startup and not using AI, you should have a damn good reason why.” and “𝗔𝗥𝗥 𝗮𝘁 𝗦𝗲𝗿𝗶𝗲𝘀 𝗔 𝗶𝘀 ~𝟯𝘅 𝗵𝗶𝗴𝗵𝗲𝗿 𝘁𝗵𝗮𝗻 𝗶𝗻 𝟮𝟬𝟮𝟭 (𝘀𝗹𝗶𝗱𝗲 𝟱𝟮) Call it a $3M median for Series A B2B companies these days. And $7M is now 75th percentile. These are not inflation adjusted, but even so the pace seems to be getting faster by the day.” 

LightSolver’s Omri Wolf to Present Keynote Speech and Participate in Birds of a Feather Panel at SCA/HPCAsia 2026.

Budibase is one of the Top 5 Retool Alternatives according to The Bulletin Time.

Aquant
Enterprise Business Development Representative (Newton MA, Hybrid)

Blue Energy
Project Controls Engineer (Great Houston, Hybrid)

Firebolt
Software Engineer (Munich)