BTS Turns AI into a Measurable Revenue Line as Token Costs Loom
Q2 2026: North America rebound, AI services hit 10% of revenue, and guidance raised as token costs are flagged for 2027.
BTS-B.ST · Earnings Call · 2026-08-14
AI Becomes a Measured Business Line
BTS Group delivered a second consecutive quarter of profit growth in Q2 2026, with revenue up 9% currency adjusted and EBITA margin improving from 11.7% to 12.3%. The more consequential development, however, is the company's decision to publicly quantify its AI-enabled service revenue. Direct AI implementation services now account for 10% of total revenue, growing 221% year-over-year to SEK 76 million. Jessica Parisi explained that these engagements are where clients explicitly ask BTS to "help us with our AI ambitions" – a shift from one-off projects to ongoing partnerships around workflow reinvention, adoption, and change management. This is not a pivot but an acceleration of a theme BTS has nurtured for several quarters. The AI innovation around simulation development and delivery has been a recurring topic, and the company has consistently described its differentiated position as an AI partner. What is new is the willingness to attach hard numbers to it. The 10% threshold crosses a psychological line and signals that AI services are becoming a core revenue stream rather than a side bet. This builds on prior commentary where Parisi described such breakthroughs as "AI diamonds" that reshape how teams operate: “we help with the global adoption and change management of the new ways of working around that diamond” — Jessica Parisi, Executive (likely CEO or CFO) · 2026-05-21 (May 2026 call).North America: Turnaround Deepens
The recovery in BTS's largest unit, North America, continues to gain traction. After a difficult 2025, the region posted double-digit organic revenue growth and improved its EBITDA margin to 37.5%. Management attributed part of the margin lift to decisive cost actions: the closure of a small unprofitable business (including $400,000 in severance) and a reduction in team size following another AI breakthrough that collapsed what used to be a seven-person design process into a two-person one, resulting in over $1 million in annual run-rate savings to be realized over the coming quarters. Parisi acknowledged that the underlying margin is even stronger when excluding one-off items. "If we actually remove those 2 unprofitable businesses in the first half and the Sounding Board one, our margin would have been up 4 more percentage points," she said (“If we actually remove those 2 unprofitable businesses in the first half and the Sounding Board one, our margin would have been up 4 more percentage points.” — Jessica Parisi, Executive (likely CEO or CFO) · 2026-08-14). This transparency helps analysts separate sustainable expansion from one-time noise.Token Costs: The Next Frontier
Perhaps the most revealing insight came in response to a question about token costs. BTS switched from ChatGPT to Anthropic's Claude suite in May, but locked in a 12-month enterprise license. That reprieve ends next year. Management estimated token costs could quadruple in 2026, rising from roughly $400,000 to between $1.2 million and $1.6 million. "We expect our token cost with Anthropic to go up 4x probably next year," Parisi said (“We expect our token cost with Anthropic to go up 4x probably next year.” — Jessica Parisi, Executive (likely CEO or CFO) · 2026-08-14). The company is betting that productivity gains from a new internal platform will more than offset that increase, but the timing will play out over the next four quarters. This is a rare instance of a consulting firm quantifying its own AI consumption, and it underscores BTS's willingness to expose its own operational exposure to the same technologies it sells to clients. As Parisi noted earlier this year, the AI technology itself carries "higher margins" on the offering side (from the February 2026 call: “With the AI technology with much higher margins on that offering, that would obviously be the other one.” — Jessica Parisi, CEO · 2026-02-20).Balancing People with AI
Employee headcount has now declined for five consecutive quarters, but that narrative has two sides. While operational back-end staff continues to shrink – a direct result of AI-driven efficiencies – BTS is actively hiring billable consultants. Parisi expects another four quarters of this "back-end/front-end balancing" before net adds resume, absent acquisition-driven growth. This is consistent with the company's capital-light model, which has historically delivered 12% revenue CAGR since IPO without additional equity funding.The guidance upgrade to "significantly better than 2025" is a direct consequence of the North America momentum and an improving outlook for the troubled "other markets" division, which is expected to return to growth earlier than the usual three-quarter turnaround. "We expect them to get to growth earlier in the second half than the end of the second half," Parisi noted (“We expect them to get to growth earlier in the second half than the end of the second half.” — Jessica Parisi, Executive (likely CEO or CFO) · 2026-08-14). BTS's strategic positioning – AI as a people and change problem rather than a technology issue – remains its core differentiator. As global enterprises grapple with AI adoption, BTS is positioned as the bottleneck-resolver. The company's way of working with clients is evolving, and the mix of consulting, licensing, and ongoing support is expected to remain stable in the near term, but the growing share of AI-specific revenue suggests a structural shift that analysts will be watching closely. The key question for the remainder of 2026 is whether the token-cost headwind arrives before the productivity savings kick in, and whether the billable consultants can scale fast enough to absorb the demand. With 1% global market share in a rapidly expanding space, BTS has ample room to grow, but execution will determine whether the AI dividend becomes a permanent part of the margin story.It means our clients are saying, help us with our AI ambitions... help us drive the adoption up... help us get our leaders to shift to create the type of environments where their teams can experiment and innovate with AI.