Engineers vs. Tokens: SaaS CEO Claims AI Costs ‘Astronomical’ at Cannes
Key Takeaways
- At Cannes Lions, SaaS and ad tech leaders reveal that AI development costs are outpacing traditional engineering, especially as token costs rise.
- The promise of 3x productivity is real, but only when AI is carefully supervised—not the unmanaged automation initially hyped.
Mentioned
Key Intelligence
Key Facts
- 1At Cannes Lions 2026, the public narrative champions AI, efficiency, and agentic orchestration, but private discussions reveal deep anxiety over rising AI development costs.
- 2Ian Maxwell, CEO of Converge Digital, stated that using AI on an entire codebase results in ‘absolutely astronomical’ costs that are ‘vastly more costly than simply having engineers.’
- 3Maxwell reports real-world productivity gains of 3x when AI augments engineers in a supervised manner, contrasting with the often-cited but unsubstantiated 10x claims.
- 4A senior marketer, speaking anonymously, noted that cost conversations are already emerging in dealings with media agencies, signaling a shift from theoretical to practical concerns.
- 5The change from a year ago is stark: Cannes 2025 focused on AI’s upside like faster production, while 2026 finds executives calculating unexpected bills from token usage and API calls.
No one really admits openly as to how much they’ve been burning to develop. ... If you throw a whole code base at it the costs become absolutely astronomical ... it is vastly more costly than simply having engineers.
Off-stage conversation at Cannes Lions 2026
Analysis
- 3x real-world productivity gains when AI augments engineers
- Avoids massive unmanaged AI costs through supervision
- Astronomical token costs when processing entire codebases
- Token price rises make AI more costly than traditional development
- 10x productivity claims largely unsubstantiated
Analysis
SaaS companies at the heart of the AI revolution are now doing an uncomfortable cost-benefit analysis, and the math is sobering. At Cannes, Converge Digital’s CEO laid bare the numbers: unleashing AI on a full codebase can be far more expensive than hiring engineers, and the claimed 10x productivity gains only hold in narrow use cases.
The 2026 Cannes Lions International Festival of Creativity has become a stage for the triumphant narrative of AI transformation, with panels and keynotes echoing buzzwords like efficiency, agentic, and orchestration. But as the rosé flows and the official sessions end, a different, more anxious conversation is emerging among advertising, technology, and marketing executives: the staggering cost of making AI work in the real world.
Ian Maxwell, CEO of ad tech firm Converge Digital, candidly told Digiday that no one openly admits how much they’ve been ‘burning’ on AI development.
Industry leaders speaking off the record and on background now admit that the once-theoretical question of AI’s price tag has become an urgent operational concern. Ian Maxwell, CEO of ad tech firm Converge Digital, candidly told Digiday that no one openly admits how much they’ve been ‘burning’ on AI development. He revealed that applying AI to entire codebases resulted in ‘absolutely astronomical’ costs that now vastly exceed the expense of simply employing engineers—a reality exacerbated by rising token costs. This marks a sharp departure from the prevailing industry myth of effortless 10x productivity gains, which Maxwell says only materialize in highly specific, narrow use cases. Instead, he reports that by using AI in a supervised, augmentative role, his firm achieves a more modest but still meaningful threefold increase in engineering productivity.
What to Watch
This reassessment is not confined to ad tech. A senior marketer, granted anonymity to speak candidly, confirmed that cost conversations are already filtering into relationships with media agencies. The shift in tone over the past year is palpable: where Cannes 2025 celebrated AI’s unbounded upside—faster creative production, leaner teams, lowered barriers—Cannes 2026 finds the same people grappling with bills they did not expect to be paying so soon. Token-based pricing models, which charge per API call or computational cycle, are proving unpredictable and can scale disastrously when ambitious AI projects are deployed without guardrails.
The implications ripple across the entire technology and marketing ecosystem. For SaaS companies building AI-powered tools, the margin pressure could be severe; for agencies promising AI-driven efficiency, the underlying cost structure may undermine the pitch. Investors and clients are increasingly likely to demand transparency, moving the AI conversation from magic to math. Looking ahead, the festival’s off-stage arithmetic will likely shape product roadmaps, with a premium on supervised, targeted AI applications that deliver measurable productivity without runaway expense. The era of unmanaged AI experimentation at the expense of the bottom line may be drawing to a close, even as the public stage at Cannes continues to sell the dream.
How we covered this story
Every story in our saas coverage is assembled from multiple primary sources, cross-referenced for factual consistency, and scored along three independent dimensions: sentiment, operational impact, and source-cluster confidence. Single-source rumors and unverifiable claims do not pass our editorial gate. When a story shows "Verified by N sources" with N≥2, the development is independently corroborated; when N=1, we mark it explicitly so readers can weigh the signal accordingly.
Impact scoring uses a 1-10 scale weighted toward regulatory, financial, and operational consequence rather than coverage volume. A topic that runs in every outlet but moves no real decisions ranks lower than a niche regulatory filing that reshapes how operators in the saas space have to behave. Read our full methodology for the scoring rubric, our glossary for term definitions, and our trends index for the longitudinal view across the beat.
| Signal on this page | What it tells you |
|---|---|
| Verified by N sources | Independent corroboration count. N≥2 is our confidence floor; N=1 is marked explicitly. |
| Impact score (1-10) | Regulatory + financial + operational weight. 8+ signals an experienced-operator action item. |
| Sentiment | Five-tier classification trained on labeled saas-specific corpora. |
| Timeline | Where applicable, the related-events sequence that contextualizes today's development. |