Google Cloud and Antler on Building AI-Native Startups in EMEA | Antler Immersion

Part 1: Market Trends & Strategic Context

1. Market Shifts

How is the AI market evolving, and why are we seeing a shift away from basic wrappers toward complex, defensible architectures?

Mariam Chahin: The first wave of generative AI was defined by speed to market, where founders built thin UI layers on top of third-party foundation models. While that was effective for testing early demand, wrappers quickly encountered a ceiling with no proprietary moat, rising compute costs, and high platform risk.

We are now firmly in the second wave of AI innovation, where enterprise buyers and sophisticated users demand systems that are deterministic, context-aware, and deeply integrated into operational workflows. The market has shifted from simple prompt engineering to true systems engineering, prioritizing multi-agent orchestration, proprietary data grounding, and domain-specific fine-tuning.

Adam French: This is a time of extreme opportunity, and extreme competition. Most of the obvious use cases for application layer AI have already been built, so founders are either looking at more and more niche solutions, or they are pivoting towards more ambitious agentic orchestration. We are seeing this in our own portfolio, and across the biggest success stories in Europe. The big funding rounds this year are being raised by frontier labs and companies that combine AI with big industrial sectors such as defence, energy, healthcare and the compute that powers them. This is where real growth now lies. 

2. Defensibility Criteria

From a venture perspective, what separates the top 1% of AI-native startups that secure funding from those that struggle to scale?

MC: The most resilient AI-native startups separate themselves through three core disciplines: workflow moats, architectural sophistication, and unit economic control.

First, they embed themselves deeply into high-friction customer workflows where daily usage creates a compounding, proprietary data flywheel. Second, their architectures rely on multi-agent collaboration and intelligent model routing rather than monolithic, single-prompt API calls. Finally, they maintain strict visibility over their token consumption, inference latency, and gross margins from day zero, ensuring that customer growth translates directly into enterprise value.

AF: Velocity. The speed at which founders can build, ship and iterate is everything in this new era. Creating a startup has always been a race to find product market fit before cashflow runs out. Now, founders can build more and get further with less as a result of AI, but that core equation hasn’t changed. How quickly can you work through as many different hypotheses as it takes to reach a viable product that people want to buy. The best founders leverage AI to get there faster than ever. 

3. Collaboration Rationale

Why are Google Cloud and Antler bringing their engineering and venture strategies together for this specific initiative?

MC: Building a generational AI company requires mastering frontier engineering and venture-scale execution at the same time. Google provides founders with direct access to frontier models, scalable cloud infrastructure, and tools like Antigravity in Gemini Enterprise. Antler brings deep venture DNA, day-zero company building expertise, go-to-market acceleration, and direct access to institutional capital.

Too often, deeply technical teams build powerful technology without a clear commercial engine, while commercial founders struggle with AI scalability. This collaboration unites both sides to give founders the complete stack needed to build sustainable, venture-grade businesses.

AF: We want to give founders the best possible chance of success. Harnessing the technology and expertise within Google Cloud gives founders leading tools and technologies. We are really excited about combining our experience as a truly global inception-stage investor with Google Cloud. 

4. Founder Ecosystem

What unique advantages do EMEA-based founders have in today’s global AI landscape, and where do they face the biggest hurdles?

MC: EMEA possesses world-class AI research hubs, dense engineering talent, and deep institutional strength across sectors like fintech, healthtech, and enterprise software. Moreover, European founders have built-in fluency with data governance, privacy standards, and regulatory frameworks like GDPR and the EU AI Act, which prepares them to deliver enterprise-ready solutions for global deployment.

The greatest hurdle for EMEA founders remains scaling velocity across fragmented markets. Navigating distinct regulatory jurisdictions, localizing go-to-market strategies, and bridging the gap between seed-stage innovation and Series A or B capital require intentional architectural agility and structured commercial playbooks.

AF: In Europe and the UK we are often our own worst enemies when it comes to discussing our innate advantages and challenges. We focus so much on the negatives that we make them our reality. The truth is that Europe has world-class founders, academic institutions and a flywheel of founders coming from unicorns that is spinning faster than ever. 

According to Antler’s own data, tech companies are now reaching unicorn status in Europe just as fast as in the US and founders have the ambition, dedication and technical expertise required to build generational companies. 

5. Venture Math & Unit Economics

Why is understanding "VC math" and unit economics critical for technical founders early in their build cycle?

MC: Unlike traditional SaaS, where software gross margins easily hover above eighty percent, AI-native startups face real infrastructure costs tied directly to inference latency, compute capacity, and token consumption. If a system is architected inefficiently, user growth will rapidly erode cash flow instead of compounding margin.

Technical founders must understand how investors evaluate customer acquisition costs, lifetime value, and payback periods in relation to their underlying tech stack. Designing with unit economics in mind from the start—such as utilizing context caching, fine-tuning smaller specialized models, and routing simpler tasks to cost-efficient models—ensures that technical choices directly support a fundable business model.

We see a similar approach in sports where elite athletes obsess over micro-metrics (exact shot angle or stroke count) to enable them to score better or swim faster. Similarly, successful AI founders must obsess over token consumption, inference latency, and gross margins rather than relying on brute-force compute.

AF: I agree with Mariam’s assessment, but would also add a specific note on ‘VC math’. Founders, whether they are generalist or technical, need to remember that the economics behind venture capital are surprisingly simple. We are looking for billion-dollar or multi-billion-dollar outcomes. Startups need to have a clear path towards that possibility to be attractive to us. 

Technical founders need to ensure they are building a scalable company, not just a good product that solves one specific problem. We see this all the time - founders identify a problem and develop a single point solution to address it. That’s a good start, but is it the foundation for a billion-dollar company? 

Part 2: Program Mechanics & Execution

6. Target Cohort

Why is the Immersion program strictly restricted to post-MVP startups with live code and customer traction?

MC: The Immersion program is an architectural and scaling sprint rather than an early-stage ideation hackathon. To extract meaningful value from Google Cloud customer engineers and Antler partners, founders need a live production environment with real code, active API traffic, and genuine user behavior.

When a team arrives with existing production systems and identified bottlenecks, our engineers and mentors can immediately step in to diagnose latency challenges, optimize agentic workflows, and refine monetization strategies in real time.

7. Technical Curriculum

How will technical leads utilize tools like Google AI Studio, Gemini Enterprise, and ADK in Phase 1?

MC: Phase 1 is an intensive digital sprint designed to transition engineering leads from simple prompt builders into full AI systems architects. Founders will use Google AI Studio and Gemini models to prototype multimodal prompts, test structured outputs, and leverage long context windows for complex data retrieval.

Antigravity in Gemini Enterprise provides an agent-first developer environment for rapid sandbox experimentation, fast prototyping, and automated agent workflows. Alongside this, teams will utilize Gemini Enterprise and Agent Development Kit to structure autonomous, multi-agent systems capable of multi-step reasoning, external tool execution, and deterministic enterprise guardrails.

8. Selection Criteria

What key metrics will Google and Antler look for when choosing the 25 startups for the Phase 2 London residency?

AF: These startups are still early in their growth journey, but they need to be able to demonstrate a high quality product and clear customer traction. There has to be demand for what they’re building, and they have to be able to sell us a vision for how they are going to scale and grow. 

Beyond business numbers, we will also be looking hard at the founders themselves. We talk a lot about founder-market fit - are they the best people in the world to be building this particular startup. What first hand experience of the industry and problem have they had? Do they have the grit and determination required to go all in and spend the next ten years building this company through all of the highs and lows that come with startup life? Where have they excelled in their lives to show they are comfortable performing consistently at the highest level under extreme pressure? 

9. In-Person Experience

What can founders expect during Phase 2’s live whiteboarding sessions and 1:1 strategy office hours in London?

MC: Phase 2 is a high-touch, full-day meet at one of Google’s London offices designed for deep technical and commercial acceleration. Founders will work directly with Google Cloud customer engineers and AI specialists in live whiteboarding sessions to refactor system architectures, reduce token expenses, and reinforce enterprise security and scalability.

AF: Founders will also have dedicated one-on-one sessions with Antler partners to stress-test everything - the problem they are solving, go-to-market positioning, pricing models,, and polish investor pitch narratives. Beyond the scheduled tracks, the residency serves as a place of exceptionally high-density talent connecting EMEA’s top one percent of AI-native founders.

10. Application Process

What is the most critical piece of advice for teams completing their applications before the October 13 deadline?

MC: Be specific, avoid generic AI buzzwords, and show your work. Clearly articulate the exact domain problem your startup solves and explain why existing foundation models cannot solve it out of the box.

Detail your current technical architecture and highlight the specific bottlenecks you are working to overcome, whether related to latency, agent orchestration, or context window management. Back this up with concrete traction metrics such as active usage, retention rates, or early revenue. We are looking for technical ambition paired with sharp commercial clarity.

AF: We are looking for founders who combine technical expertise and great storytelling. We need to know the technology works, but that founders also have the vision to sell this to customers, to attract great talent and to secure investment.