Welcome back to another episode of Upside where Dan Bowyer and Mads Jensen of SuperSeed go behind the headlines shaping European tech, capital, and power.
This week is an AI-heavy sprint with a guest who’s right in the Gulf capital flow: Sam Marchant. Anthropic’s monster round is the headline, but the more interesting story is underneath: enterprise AI is becoming workflow-sticky, while OpenAI feels like it’s drifting toward consumer monetization experiments.
Then we get into the “AI productivity” paradox: why generative tools aren’t giving us leisure, they’re giving us more output… and more work. From there: Alphabet’s 100-year bond and what it says about tech becoming a utility, plus the uncomfortable European angle — our savings funding US hyperscalers while we debate sovereignty.
Finally, Europe sovereignty vibes: Mistral’s enterprise ramp, the 28th regime rhetoric, and whether political systems can actually execute. We close with space: Orbex collapsing, “data centers in orbit,” and why maybe civilization needs billionaires burning capital on high-variance cathedral projects.
This is Upside, where optimism is earned, not assumed.
What’s covered:
00:21 Anthropic’s $30B: why the market can’t stop throwing money at enterprise AI
03:42 The real shift: OpenAI → consumer/ads vibes, Anthropic → coding + enterprise execution
04:50 Gulf capital dynamics: OpenAI relationships vs QIA showing up in Anthropic
07:21 Claude vs ChatGPT: switching costs are collapsing… until workflows become the moat
10:54 HBR’s “AI intensifies work”: why productivity becomes pressure, not leisure
12:19 Autonomy + mastery + dopamine: AI as the ultimate short feedback-loop machine
13:25 Practical use cases: research across languages, idea stress-testing, “AI as a first hire”
22:05 Alphabet’s 100-year bond: tech is now priced like infrastructure
24:51 The pension problem: Europe’s savings financing US scale while Europe underfunds Europe
32:44 Europe’s GDP gap is a tech gap: productivity isn’t the issue, tech scale is
39:51 Mistral’s enterprise ramp: sovereign AI or local services + transformation advantage?
45:37 The 28th regime: big words, hard execution — can Europe actually push reform through?
50:32 Space data centres: PR-on-steroids or physics-defying inevitability?
53:07 Orbex collapses: why “mid-sized countries” can’t win launch alone
55:20 Fusion/quantum: Europe’s deep R&D edge, blocked by capital markets structure
56:25 Deal of the week: Olex’s $1B+ moment and Europe’s chip-shaped ambition
🎧 Listen on Apple Podcasts or Spotify and if you’re building or investing in AI, this one is worth queueing with chapter markers.
Anthropic’s $30B: The Enterprise AI Flywheel
Anthropic’s raise isn’t just “big number go up.” It’s a signal: investors are treating the enterprise AI category like it’s already won and Anthropic is positioned as the cleanest winner.
What matters here:
Coding + enterprise are the money engines.
Claude Code is sticky in a way chat UIs aren’t.
Once dev teams build workflows, ripping it out is painful.
That’s the real moat: not “model is better this week,” but “your org is wired around it.”
OpenAI is still massive, but the narrative divergence is real: Anthropic feels like execution; OpenAI feels like strategy pivots.
Gulf Capital: The Next Quiet Proxy War
Sam’s point from the region is sharp: in the Gulf, AI has to create upside, not just cost savings.
Because labor structures and cost bases don’t always make “efficiency” the killer pitch, the winning deployments are the ones that can say:
revenue expansion
new value capture
strategic differentiation
Also: watch the capital pools.
If OpenAI is “the relationship bet” in the UAE, and QIA shows up in Anthropic, you can see how this becomes a soft rivalry of influence: different sovereign pools backing different AI stacks.
Claude vs ChatGPT: Switching Is Easy… Until It Isn’t
The most telling moment: everyone casually admitting they’ve moved to Claude.
Because for most people, switching costs are basically zero.
You follow the juice.
But that changes when:
you’re in terminal workflows
you’ve built automations
you’ve embedded it into how you operate
That’s why Claude Code matters. It’s not “better answers.” It’s “better habits.”
AI Intensifies Work: Productivity as Acceleration, Not Leisure
HBR’s framing lands because it matches lived reality:
AI doesn’t reduce work. It increases what you can do.
And if your goals stay the same, you get time back.
But if your ambition expands (it usually does), you just do more.
Mads’ lens is the best one:
Autonomy: you don’t need anyone else to ship
Mastery: you can operate across domains
Purpose: you feel like what you’re doing matters
That combo is… rocket fuel.
Work becomes addictive because feedback loops collapse to minutes.
Alphabet’s 100-Year Bond: Tech as a Utility
A 100-year bond is a statement:
Alphabet is being priced like infrastructure.
But the European angle is the uncomfortable one: this is exactly where European savings can end up — long-duration, “safe” assets that quietly underwrite US hyperscaler dominance.
Which connects to the deeper issue:
Europe doesn’t have a “lack of capital” problem.
It has an allocation architecture problem.
Pensions, mandates, and market structure push capital into safety optics instead of long-term growth engines.
Europe’s GDP Gap Is a Tech Gap
The cleanest framing:
Europe’s prosperity gap isn’t “hours worked.” It’s tech scale.
If you strip out tech, Europe looks fine.
If you include tech, the US pulls away.
So sovereignty isn’t just defense. It’s economics:
No tech scale → weaker productivity growth → weaker fiscal capacity → weaker state.
That’s the loop.
Mistral: Sovereignty, Service, and “Local First” AI
Mistral’s ramp is the most interesting European AI story precisely because it isn’t just API revenue.
It’s enterprise contracts, transformation work, proximity, and trust.
If you’re a major European bank or industrial:
you don’t just want a model
you want a partner
you want fast iteration
you want the data story clean
you want someone in your time zone with a SWAT team
Sovereignty is a tailwind — but so is local service.
The most likely outcome is not “Mistral beats US models.”
It’s “Mistral is good enough + closer + safer + embedded.”
That’s a real business.
Orbex, Space Data Centers, and Funding the Irrational
Orbex filing is grim, but it’s also the recurring European lesson:
Mid-sized countries can’t win certain games alone.
Airbus worked because Europe pooled.
Launch will require the same logic.
And then there’s the fun-but-unsettling corner: “data centers in space.”
Physics is brutal:
cooling
radiation
maintenance
payload economics
failure rates
But Andrew McCallep’s point is the one that sticks:
Civilization advances when people fund high-variance cathedral projects.
We should want more billionaires burning capital on weird frontier bets — not cosplay luxury.
Deal of the Week: Olex
A 25-year-old British founder, James de Combe, building an AI chip company valued north of $1B is the kind of signal you want to see.
Even if Olex doesn’t win, it matters that:
someone is attempting it
in Europe
at scale
with ambition
Nothing changes until someone decides to build.
One-Line Takeaway
AI is rewriting the workday, hyperscalers are being priced like nation-states, and Europe is talking sovereignty, but the whole thing still bottlenecks on whether we can fix capital allocation and execute reform at speed.
