Lovable's $400M Ramp Needs 150 ML Hires. Stockholm Can't Supply Them.

On August 12, 2026, Lovable closed a $400M Series C at a $13.3B valuation and said it plans to reach roughly 450 employees, with the heaviest hiring in ML, infrastructure, and security. If you're the recruiter or founder who has to actually fill those seats, the raise is not the story. The story is that Stockholm's ML pool is a few hundred engineers total, spread across every employer in the city, and Lovable wants a meaningful slice of them.

This piece is about the arithmetic Lovable's press release skipped.

Why Lovable's five-city footprint is a sourcing map, not a growth map

Lovable is headquartered in Stockholm, but the Series C announcement names London, Boston, San Francisco, and New York as expansion cities. That is not global ambition. That is forced sourcing, because Stockholm cannot supply the engineers on its own.

Two constraints make the math obvious:

• Glassdoor's Stockholm feed lists roughly 271 ML-titled roles open across all employers as of June 2026, up from 248 in March. Across Sweden as a whole, the number is 444 as of August 2026.
• Lovable's own listings on Wellfound show 25 open engineering roles across Stockholm, SF, UK, and US, meaning a single 11 to 50-person company is already carrying close to 9% of Stockholm's visible ML demand.

You cannot hire 150 ML and infra people from a city where the total employer-wide demand is under 300 seats and Spotify, Klarna, King, and Modal are all fishing the same pond. The London, Boston, SF, and NYC offices exist because the candidates already sit there.

3.2:1

The mechanism behind the split

Lovable's revenue is running at a $500M annualized rate as of June 2026, projected to end August at $600M. It also signed a multiyear Google Cloud deal that increased compute usage fivefold, and it trains its own in-house model alongside frontier options. That means the infra org is not a "keep the lights on" function. It is core product, supporting 60 million projects and 900 million monthly visitors. Core product headcount does not tolerate a two-year Stockholm-only ramp when the pool is this thin, so the money goes to relocation, US comp bands, and satellite offices near existing talent density.

The real Stockholm ML pool, and where it actually lives

The honest read: Stockholm's senior ML and MLOps population is small, and the majority sits inside a handful of employers. Any sourcing plan that ignores that concentration is wishful thinking.

The likely poach targets, ranked by realistic yield:

1. Spotify (3,200+ Stockholm R&D employees, per KiTalent). The default academy. Expect the ML Research and Personalization groups to bleed first.
2. Klarna. Payments fraud ML and MLOps at scale. High overlap with Lovable's infra needs.
3. King. Game telemetry ML, ranking, and matchmaking. Adjacent skills.
4. iZettle alumni network. Fintech ML, dispersed but locatable.
5. Modal. Actively hiring a Forward Deployed Engineer - ML in Stockholm, per Glassdoor. Direct competition for the same handful of candidates, and a source of engineers who already said yes to a small AI company once.

Naming the source companies is the easy part. Finding the specific individuals inside them who work on the systems Lovable actually needs (retrieval, inference optimization, training infra, guardrails) is where keyword sourcing on LinkedIn falls apart. Titles like "Senior Software Engineer" at Spotify tell you nothing about whether the person spent three years on vector recall or three years on iOS builds. This is the exact gap Refolk closes: you describe the person in plain English ("Stockholm-based ML engineer, worked on retrieval or ranking infrastructure at Spotify or Klarna, comfortable with GPU inference") and get a ranked shortlist grounded in GitHub, LinkedIn, and the open web.

The comparable figures, in one table

Here is the dataset in one place, so you can argue with your CEO from the same sheet of paper.

Two rows to sit with. Lovable is a single 11 to 50-person company carrying ~9% of the visible Stockholm ML demand. And Sweden's broader IT and cybersecurity shortage exceeds 30,000 people, which matters because Lovable is also hiring Trust and Safety engineers in Stockholm right now.

Why the 5-days-onsite rule is the single biggest sourcing constraint

Lovable's public infra job description says candidates must be "based in Stockholm or ready to relocate - this is an on-site, 5-days-a-week role." That one sentence is doing more damage to the hiring plan than any comp band could.

It excludes, in order of severity:

• The remote-first Nordic ML population that scattered to Copenhagen, Oslo, Berlin, and Lisbon over the last few years.
• The London ML pool, which is larger than Stockholm's and unwilling to trade a London salary for a Stockholm one that KiTalent describes as "structurally discounted" against London and US remote comp.
• EU relocators who are willing to move for a later-stage round but not a Series C.
• Ex-US-remote engineers who specifically will not go back to five days.

Lovable will pay to relocate or lose the search. There is no third option in Stockholm.

If you are sourcing for Lovable or for anyone competing against them, the practical read is that Stockholm hires will skew ex-Spotify, ex-Klarna, ex-King, ex-iZettle, plus a slow trickle of EU relocators bought with equity and relocation packages. Everyone else has to be recruited into London, Boston, SF, or NYC, where remote and hybrid are still on the table.

Security hiring is the hidden bottleneck, not ML

Security is where the plan quietly breaks. Sweden's IT and cybersecurity shortage is over 30,000 people, security engineers are harder to relocate than ML engineers, and Lovable is hiring Trust and Safety in Stockholm on top of the ML and infra roles.

Two reasons security is thinner than ML in this specific market:

• Context transfer costs. A security engineer who knows your threat model, your cloud posture, and your incident playbooks is worth three generalists. Relocating that person means paying for the year of context they'll lose.
• Regulatory friction creating a brand-new category. EU AI Act enforcement is phasing in through 2026, with compliance costs of €400,000 to €2.5M for high-risk providers per AI Sweden's Regulatory Impact Assessment. That has created a brand new "AI governance and compliance engineering" category that barely existed two years ago. There is no trained pool.

Where the non-Stockholm 150 will actually come from

Lovable's London, Boston, SF, and NYC hires will come warm through the investor network first - Menlo Ventures led the round, and the Scaleup Europe Fund managed by EQT co-led - then through targeted poaches from regional employer clusters. That is the pattern every Series C at this scale follows, and it is more predictable than the press releases suggest.

Named enterprise customers Lovable can lean on for warm outreach - Nvidia, Adidas, Hearst, and Zendesk, per SiliconAngle - matter here too. Engineers who have been on the buying end of Lovable's product at a large enterprise are pre-qualified believers, which is a real sourcing edge if you can locate them. A prompt like "engineers who have publicly talked about integrating Lovable at Nvidia, Adidas, Hearst, or Zendesk" is exactly the kind of open-web question Refolk was built to answer, because it needs GitHub commits, conference talks, and LinkedIn posts stitched together, not a Boolean string.

The GPT Engineer GitHub community is worth naming here as well. Lovable emerged from that open source project - founders Anton Osika and Fabian Hedin started the company in late 2023 - and the contributor graph is a legitimate sourcing channel that most recruiters ignore.

What sourcers should actually do this quarter

Stop treating Lovable's five-city expansion as five separate searches. Treat it as one global search with four regional filters, prioritized by pool density and 5-day-onsite tolerance.

The playbook, in order:

1. Map the Spotify, Klarna, King, and iZettle-alumni Stockholm graph first. These are your only realistic on-site hires at scale. Assume a 60-day cycle.
2. Open London second, remote-friendly. London ML supply is larger than Stockholm's, and comp expectations sit between Stockholm's discounted band and US remote.
3. Run Boston and NYC in parallel. Both are quiet enough that a well-targeted outreach cadence lands.
4. Save SF for the roles where nothing else works. SF is where you spend the last of the round on comp, not the first, because you are bidding against Anthropic, OpenAI, and every other vibe coding startup for the same candidates.
5. Tag every candidate by AI Act exposure. Compliance-literate ML engineers are the new scarce category. Filter for it at sourcing, not at interview.

The engineers exist. The pool is thin but not zero. What is missing in most sourcing stacks is the ability to describe the person you want in plain English and get a ranked shortlist back, which is why I built Refolk. Boolean search cannot express "ex-Spotify Personalization, shipped retrieval infra, publicly relocatable to Stockholm." A sentence can.

FAQ

How many ML engineers are actually in Stockholm?

Public data does not give a clean total, but Glassdoor's Stockholm ML job feed shows 271 open roles across all employers as of June 2026, and Sweden as a whole shows 444 as of August 2026. The senior, hireable-in-the-next-90-days pool is a fraction of that. Spotify's 3,200+ Stockholm R&D headcount is the largest single reservoir, followed by Klarna, King, and the iZettle alumni network.

Why is Lovable opening London, Boston, SF, and NYC offices if it's Stockholm-based?

Because it cannot find 150 ML, infrastructure, and security engineers in Stockholm alone. The five-city footprint is where the candidates already sit. London supplies senior ML at scale, Boston supplies infra-heavy engineers, SF supplies the frontier AI cluster, and NYC supplies applied AI and quant-adjacent ML. It is a sourcing map more than a growth map.

Is the 5-days-onsite Stockholm requirement realistic given the pool?

Only for a subset of hires. Lovable's own infra job description specifies Stockholm on-site five days a week, which excludes remote-first Nordic ML talent, the London pool, and most EU relocators unwilling to move for a Series C. Expect Lovable to either soften that policy for senior hires or lean hard on relocation packages, especially for ex-Spotify and ex-Klarna candidates already in the city.

Where is the hidden bottleneck?

Security, not ML. Sweden's IT and cybersecurity shortage exceeds 30,000 professionals, security engineers are harder to relocate than ML engineers, and the EU AI Act's phased 2026 enforcement has created a brand-new "AI governance and compliance engineering" category with no trained pool. If Lovable's ramp stalls, it will stall on Trust and Safety and compliance seats before it stalls on ML.