Hello everyone, and thank you for watching. I’m Tange, President and CEO of SHIFT. Thank you for joining our live earnings briefing for the third quarter of the fiscal year ending August 2026. Today, I’ll walk you through our third quarter results, the progress of our sales activities, and our initiatives for the Native AI era. We’ll then move on to the Q&A session. For questions, please enter the chat with your institution name and your name. We’d also appreciate it if you could submit your questions before our presentation ends. Now, let me begin. First, our earnings highlights. In the third quarter, revenue grew about 20% year-on-year, and adjusted operating profit rose about 23%, very solid growth in both revenue and profit. Gross margin, however, was soft at 32.5% and 32.8%, and I’ll explain why shortly.
I’d also highlight that AI-related revenue reached about 19% of total revenue, and AI gross margin came in at 43%, a very strong result. On gross margin, our sales activities delivered strong results. Sales headcount rose only about 20, from around 161 in last year’s first quarter to 181 today. Our priority was to build the sales system itself, so we deliberately grew our capability with this lean team. We kept building that capability while also focusing on AI-driven revenue and on our project unit price and revenue per client. By focusing on AI revenue and high-value, long-term projects, we’ve steadily built future sales and won larger projects. As a result, utilization dipped temporarily in the third quarter. Since July, though, both project volume and utilization have recovered strongly, and we expect to raise utilization in the fourth quarter.
Our future revenue pipeline now stands at about JPY 75.4 billion, roughly 1.5 x last year’s level. For Native AI, what was only about JPY 70 million in last year’s third quarter has grown into a future pipeline of about JPY 7.6 billion. We’ve steadily built up future revenue. On SG&A, as I’ve mentioned, we’ve driven efficiency through in-house AI BPaaS and lowered the SG&A ratio. We also revised our hiring policy, and the effects are starting to show. With progress on fixed cost efficiency too, the ratio, excluding the Nisseicom M&A, came to 19.7%, a meaningful reduction. Turning to our revised forecast, we initially guided revenue of JPY 150 billion. During the period, Nisseicom joined the SHIFT Group, and with the M&A effect, we’re revising revenue up to JPY 160 billion. Adjusted operating profit is unchanged. Unfortunately, on our investment in Rise Consulting, we’re recording an impairment.
Accordingly, we’re revising adjusted recurring profit to JPY 16 billion and adjusted net income to JPY 9 billion. As we flagged earlier, an impairment impact was expected, and these are now the finalized figures. On PMI, over the past year, we’ve worked on our large investment deals, Nisseicom and Rise as a minority stake. PMI is progressing well. At Nisseicom, we’ve begun changing the cost structure, seconding sales and back-office staff to SHIFT to reduce costs. Those efforts are already underway. With Rise, our stake was about 30%, so collaboration had been limited. But demand in consulting is very high, so we’ve stepped up the partnership, working far more closely on hiring and sales support, and beginning a deeper alliance. On M&A strategy, SHIFT has historically been an active acquirer.
Given today’s environment, we want AI firmly at our core, so our top priority is organic growth. Within that, we’ll concentrate investment on AI and on improving gross margin. We’re committed to investing in organic growth. For M&A, we’re building an M&A investment platform. Using external capital and leverage, we’ll go beyond our own funds and drive M&A, starting from outside capital. We’re adjusting our capital allocation for the AI era. Since we’ll fund M&A with external money, internal cash will naturally free up. Our policy is to strengthen the once-impaired balance sheet. We’ll direct about 80% of the cash flow we generate toward rebuilding it. The remaining 20% or so, we plan to return to shareholders. Of that 20%, we plan about 10% for dividends and 10% for flexible measures, potentially share buybacks or cancellations. On that basis, we’ll begin dividends in 2026.
With net income of about JPY 10.5 billion as the base, roughly 10%, or about JPY 1 billion, will fund our first dividend. We’ll start with dividends and make sure we properly return value to our shareholders. Next, our sales progress. Over the past year or two, we’ve steadily advanced sales reform, and the results are now clear. As the chart shows, year-on-year growth keeps rising, about 30% on a consolidated basis. Our AI revenue mix has grown as well, both With AI and Native AI. In last year’s second quarter, that was about JPY 2.5 billion. This third quarter, it reached JPY 7.5 billion. We’re firmly generating AI revenue. Here are our structural revenue KPIs. As I’ve said, the key indicators for us are the monthly project unit price and the number of projects, and both keep rising each quarter. Revenue per client is climbing, too.
Healthy KPIs overall. I’d especially point to why the monthly project unit price rose. We’ve long been strong in small SES-style projects, but we’re working to become more than a system integrator, a service integrator that adds to our clients’ profits. To do that, we needed to win mid-size and large projects, not just small ones. Now, large projects, which we define as JPY 30 million or more per month, have grown to about 20% of revenue. Looking at average monthly project unit prices, small projects run about JPY 3 million a month, while large projects have grown to between JPY 50 million and JPY 66 million, up 134% year-on-year. We’ve clearly built the structure, sales strength, and delivery capability to win large projects. Our sales team’s underlying strength has grown.
As the expansion and contraction rates below show, even when a project ends, we quickly backfill with large and mid-size projects. Expansion is rising, contraction is shrinking, and our repeat rate is climbing. Very encouraging results. The contraction rate itself has come down, and overall, our repeat rate is rising, which we see as very positive results. This shows our AI revenue trend. As I noted, AI revenue jumped from about JPY 2.5 billion in the second quarter to JPY 7.5 billion in the third, a dramatic increase. The key point is that across our segments—consulting, development, testing, and BPaaS—we’ve built the service lineup and sales capability to generate both Native AI and With AI revenue in each phase. In short, we’ve now established that service lineup and sales structure. On our sales structure, we’ve built the system, so let me explain
Even at this revenue scale, we have only about 181 salespeople. We wanted to raise each person’s productivity, so rather than adding headcount, we lifted productivity. As a result, annual revenue per salesperson is JPY 0.56 billion, about 3x the roughly JPY 0.2 billion typical of an SIer. To break it down further, our salespeople fall into distinct roles: top approach reps, account reps, and follow-up reps. For top approach reps who win large projects, we can now both develop them in-house and hire from outside. The numbers are still small, so we want to grow this group. We also want more account reps for mid-size projects. To capture small projects, we’re strengthening follow-up sales. Strong follow-up selling lets us fill the gaps around large projects and do the ground-level work that lifts utilization, repeatably, so we plan to expand this team. The system is in place.
We have about 98 today and aim to roughly double that next year. Each representative generates JPY 0.4 -JPY 0.5 billion a year. A follow-up representative, on about JPY 8 million in salary costs, roughly JPY 4.6 million to hire. Generating JPY 0.4- JPY 0.5 billion, they produce about JPY 48 million in profit, for an LTV of around JPY 0.6 billion. That’s JPY 0.6 billion against JPY 4.6 million, about 130 x. It’s a highly effective growth lever, and we’ll keep hiring to make it repeatable. Next, I’d like to report on consulting in the Native AI era. Consulting is a growth area. Broadly, comprehensive consulting and specialized operational consulting together represent about JPY 2 trillion in Japan. Within that, we provide IT consulting and PM support, and we’re determined to capture this space. So why can SHIFT win here, and why is revenue growing?
As the chart shows, consulting revenue has risen sharply this year. The reasons are the four cases shown here, case one through case four, which give us a structure built to grow revenue. Case one: Traditional consulting is often simply too expensive, or clients feel they could find alternatives elsewhere. Because we offer cost-effective consulting, this switching pattern resonates strongly with clients. Case two: We began in testing, then development, building deep frontline know-how. Today, we handle RFPs, and our consulting skills have risen from downstream to upstream, so we can now win these projects. Case three: In this JPY 2 trillion market, large enterprises turn to the big four firms and similar. Smaller companies often can’t afford that and want cost-effective consulting like ours. Serving those small and mid-size clients has resonated well, and it’s driving revenue growth there, too.
Within consulting itself, one person can now handle several projects by using AI solutions, improving efficiency. Clients also want to build AI-based services, and we see strong demand for the AI-enabled consulting we provide in response. Our monthly revenue per consulting client has also risen steadily every quarter. That’s because our own consulting skills have grown. We can now build analytics dashboards, draft RFPs, design systematization concepts, and drive process improvement, so revenue per consulting client keeps climbing. Today, we provide consulting to about 230 companies. With about 3,800 active clients, we’re far from offering consulting to them all. That leaves room for roughly 3,600 more, so I believe the potential here is still very large. Next, AI modernization. AI modernization is steadily growing revenue. Our order backlog has now reached about JPY 2.83 billion.
We began this only about six months ago, and I’d like to give three main reasons for the strong sales, some of which overlap with what we shared in the second quarter. First, our mid-tier clients. Major SIers often lack the capacity, and midsize vendors often lack the skills, so our AI modernization reverse engineering service, which visualizes the system, has been very well received, and we win it without competitive bids. We also offer a product called DQS for RE. With it, clients can begin with maintenance without building a new system, going straight from reverse engineering to maintenance. It’s been very well received, and we’ve already won about 10 projects. Traditionally, SIer maintenance estimates were unclear, and clients got locked into a system. Their system’s inner workings had become a black box, so it was unclear what to fix or maintain.
Our visualization tools make that clear, delivering real cost savings and reliable maintenance, very well received. As we noted in the second quarter, in the public sector, our strong technical skills and cost efficiency are winning competitive bids, too. We’re winning those bids consistently, with a solid track record of securing these projects. On AI modernization, we’ve been at this about five months. In the second quarter, I reported 100 million steps. Three months on, we’ve reached 200 million, roughly twice the source code entrusted to us. I think that shows how well our source code visualization tool resonates and how squarely it hits our clients’ problems. On proposal opportunities, last time we had JPY 7.4 billion, and we’ve grown that only about 1.5x , so some of you may worry the pace is slowing.
This service is still new, and we’re still exploring, testing different targets, clients, and needs. Because it’s still early and exploratory, the opportunity figure has come down. As I showed earlier, our targets are now clearer: small and midsize firms going from reverse engineering straight to maintenance, and public projects we can reliably win, so they’re much easier to define. Of the JPY 11.1 billion opportunity, we’ve booked JPY 2.83 billion. Growth may look slow, but part of that reflects the timing of strong public wins in the second quarter. Being able to close sales from that JPY 11.1 billion is a positive sign, and we’ll spend the next six months or so refining our conversion rate. This page explains how. Happily, clients are now entrusting us with their source code. The people winning this work are our highly skilled ST members who design the modernization strategy.
The challenge is that we have only six ST members, and that’s become a bottleneck, limiting how much we can propose and win. Each one generates about JPY 1 billion in revenue a year, so growing this team from 6- 20, then 30, is our single most important task ahead. Growing this team from 6- 20 and then 30 will be our most important initiative going forward. Next, testing. We’re generating solid revenue here, and I’d note that AI is gradually changing how the work is done. In case one, clients are shifting toward in-house development, in a good way. AI now lets them build more themselves. As a result, systems produce more code, so QA volume rises, and more of that testing is outsourced to SHIFT. AI is especially effective for unit testing, checking code right after it’s written to confirm the system runs correctly.
User-side testing, UAT, is much harder to automate With AI, and clients often lack the resources for it. That is where we focus. We provide UAT support and use AI to expand coverage into areas we could not reach before. It has been very well received, and we are winning significant revenue there. Thanks to our brand, we are also seeing growing demand to apply AI to company-wide quality management on large projects. That kind of large-scale quality assurance is translating directly into testing revenue. This is our AI approach to testing. Through the second quarter, we focused on Nemuranai, which uses AI to automate test execution. It has already generated about JPY 0.2 billion, and in the web domain, Nemuranai has proven highly effective. We will keep expanding it, and on June 1st, we began using AI in test design.
Test design is the thinking part, deciding what to test, how to cover it, and how to assure quality. It delivers results regardless of the client’s environment, whether hardware, mobile, or web. Launching in test design was a real highlight. Look into the fourth quarter and next year, we plan to steadily widen where we apply AI across testing. Now, AI BPaaS. As I explained in the second quarter, there are two initiatives. First, across SHIFT and our group companies, we are fully AI-enabling a back office. Second, we launched an AI BPaaS division to become a profit center, winning client projects and generating profit. First, at SHIFT alone, we have driven efficiency across our 987 back-office staff using AI. Those 987 fall into three layers.
The revenue-generating salespeople, about 180, as I mentioned, plus around 60 in marketing and strategic product sales, form a strategic sales team of roughly 240. Because these people drive revenue directly, we are actively growing the team. Separately, we need about 460 back-office staff as essential infrastructure, and we will maintain that group. AI has the biggest impact on our other operations. There, we cut headcount from 284 - 174 by year-end, removing about 110 temp and business partner staff. Moving toward AI BPaaS, we aim to cut roughly another one next fiscal year. This list shows the details, which of the roughly 110 roles were reduced, and by which AI solutions. Within AI BPaaS, this is where we aim to generate revenue, mainly the IT systems area and new areas that now include sales support. We split these into a blue area and a red area.
In the blue area, our own services, WASURENAI and NAKUSANAI, are the entry point, letting us embed deeply with clients’ IT systems departments. We handle PC setup, security setup, monitoring, and helpdesk outsourcing, and we have booked about JPY 2 billion in orders. We want to grow profit here further through AI. We have worked on this for about three years, and this year, under AI BPaaS, we spent the past year on the pink area. Our FDE members, who embed AI at clients, have been active since January, already delivering about JPY 0.38 billion across 35 clients. In the IT systems area, we start by selling PCs in small lots, then win IT systems operations outsourcing. From there, we have moved into providing helpdesk, SOC, and robust security monitoring. On FDE, we built our own agent called Tensai-kun.
On FDE, we have built about 4,400 of our own AI agents called Tensai-kun. By accumulating AI know-how through them and having our FDE team embed directly in clients’ operations, we achieved about JPY 0.38 billion in revenue. FDE follows clear steps. First, we help clients raise their AI usage, then improve their productivity, and then support AI development in specialized fields. Being able to win clients in such a balanced way has been a real positive for us. Finally, to wrap up, in the second quarter, we shared SHIFT’s AI strategy. We expect the JPY 16 trillion DX market to keep growing, shifting from the traditional person-month model toward With AI. On top of that, With AI BPaaS, we’re targeting a JPY 40 trillion market beyond systems, capturing it efficiently With AI. Beyond systems, we aim to capture that JPY 40 trillion market efficiently using AI.
It’s still early, and 2026 is really year one for AI. For With AI, we already project future revenue of about JPY 16.5 billion, and within that, high-margin Native AI accounts for a solid JPY 2.8 billion. The question is how to grow this, balanced with our traditional human-led business, delivering knowledge resources through both people and AI. Whatever era comes, we’ll stay balanced. As the AI era advances, our gross margin naturally rises, and we’ll combine our hiring strength with our AI-building strength. By combining our recruiting strength and our AI-building strength, we’ll deliver our full capability to clients and grow both revenue and profit. Finally, the fourth quarter still remains, but for fiscal 2026, we expect to land at about JPY 160 billion in revenue and a gross margin of around 33.5%. Adjusted operating profit should reach about JPY 20 billion.
With one quarter to go, we’re determined to deliver. Looking further out to fiscal 2027, from that JPY 160 billion, with continued organic growth on the sales foundation we’ve built, through both everyday project sales and larger project wins, we believe we can reach about JPY 190 billion. On gross margin, with better utilization, disciplined project management, and a higher AI revenue mix, we can deliver 34.5%, even at the low end. As this progresses and the AI mix rises, we’re confident we can reach a 35% gross margin. On revenue, using our M&A investment platform, we believe JPY 190 billion could climb toward JPY 200 billion. We’re steadily delivering results. Through AI-driven back-office efficiency, we can bring the SG&A ratio below 21%, with roughly 20% efficiency gains.
For fiscal 2027, even at the low end, we aim to secure a 13.5% operating margin and JPY 25.5 billion in operating profit, and we’ll build the structure to get there. That concludes my third quarter report. We see SHIFT as a company that solves social issues by addressing population decline, labor shortages, and the technological change AI brings, and turning them into solutions, we want to contribute to society. We hope you'll continue to support the SHIFT Group. Thank you very much for joining SHIFT's third quarter earnings briefing today. We truly appreciate your continued support. Now, we will move on to the Q&A session. Either myself or Kobayashi will be answering your questions. Here is the first question. While you plan to increase the number of sales staff, will you continue to focus on high-end engineers as you have in the past? Kobayashi will answer this question.
As mentioned in the presentation materials this time, our sales structure has become quite well organized. Therefore, we will continue to hire for sales positions with the aim of increasing our top line and improving our utilization rate. Through this, both our workforce and sales will grow. As we have communicated before regarding the hiring of engineers, we plan to focus on hiring high-end talent and people who can handle project management. Additionally, by advancing AI integration, we will also work to increase productivity per person. So rather than simply increasing headcount just because sales have gone up, we intend to carefully select the talent needed for the AI era and have them join the SHIFT Group. That's the kind of approach we're aiming to establish.
Moving on to the second question. In Q2, you indicated that operating profit would be around JPY 5 billion, but the actual results were significantly lower. What was the reason for this substantial shortfall? Kobayashi will also answer this question.
Regarding Q3, for us, March, April, and May are periods that are heavily affected by seasonality. In that context, as mentioned in the materials, we focused our sales activities on long-term projects and large-scale projects, and have been working to promote AI adoption for the future. At the same time, we placed less emphasis on small-scale seasonal projects that are seasonal in nature to supplement our current utilization rate. As a result, our utilization rate declined, and we weren't able to sufficiently increase it. This is the reason why our utilization rate ended up being lower than we had forecasted. Regarding this, as shown in the materials, since July, we've been able to make adjustments on the sales side so that our usual utilization rate measures are in place, just like before. We intend to steadily work on improvements from here on out.
All right. This is the third question. Regarding With AI and Native AI, are you able to set contract prices without being constrained by human labor rates? Please explain the current situation and future outlook, including performance-based or value-added pricing. Kobayashi will also answer this question.
First, regarding the definitions of With AI and Native AI, for With AI, as before, we are working to improve productivity by using AI in areas where people are working, thereby increasing productivity, and in turn, improving our gross profit margin. This approach is less about pricing based on man-month unit costs and more about providing our services to customers as usual, charging for the work as one man-month per person. Within that, for example, by increasing productivity and finishing tasks more quickly, we can deliver the same value and thereby raise our gross profit margin. Also, in the past, you had to accumulate experience, like you wouldn't reach a certain skill level unless you worked for five years. But now, thanks to AI, you might be able to achieve that level of proficiency in three or even two years.
In this way, by using AI, we are increasing our profit margin by boosting productivity. Also, in our Native AI, revenue is generated primarily through the value of the deliverables produced rather than through human resource allocation. So when it comes to Native AI, we're able to achieve profit margins on a completely different level than before. As we continue to roll out these services, we're also working to increase the sales ratio of Native AI.
Moving on to the fourth question. Regarding AI token costs, what kinds of LLMs are you currently using? If you have any comments regarding the management of token costs or their relationship to pricing, please share them. Kobayashi will also answer this question.
First, regarding AI, company-wide, we use Copilot, and we also use various LLM models like Gemini and Anthropic. However, each LLM has its own strengths. For example, when it comes to research or brainstorming ideas, one LLM might be better, while for tasks like creating documents or analysis, another might be more suitable. Because of that, the token cost can vary significantly, so we closely monitor token consumption and associated costs on a daily basis. We are currently considering strategies for how to allocate usage as well. That is why I believe that managing token costs will become extremely important going forward. We are also focusing on making sure AI is used efficiently, not wastefully, and that it is applied to thoroughly boost productivity. We see this kind of management as crucial, and we are actively working on it now.
We have received a fifth question. What are the conditions for achieving next fiscal year's business performance? Specifically, what conditions need to be met to realize improvements in gross profit margin for both the bottom and challenge targets? I will answer this question. When it comes to gross profit margin, it is actually quite straightforward. Utilization rate has a significant impact. Also, regarding the proportion of AI revenue, we handle both Native AI and With AI. Since the gross profit margin for AI is higher than usual, the proportion of that revenue will be one of the factors contributing to an improvement in our gross profit margin. Additionally, although we have not officially announced this, whether the unit price for consultants and engineers increases in small increments could also have a slight impact. In that sense, I think those three main conditions will have a significant influence on everything.
Also, to improve bottom-line profit, it is important for gross profit to rise and for the ratio of selling, general, and administrative expenses to decrease. We want to steadily lower the SG&A ratio as much as possible. Right now, I believe around 21%. For the SG&A ratio, we are implementing a shared office system across the entire g roup company, and even with about 1,600 back-office members, we plan to maintain this structure as we aim for sales of around JPY 300 billion. In that sense, this is our target for improvement, but we hope to reduce the SG&A ratio by 1 point each year. We want to build a structure that can reliably generate operating profit. This is the sixth question: What percentage of the token cost is accounted for by sales staff? Are there any plans to disclose this information in the future? Kobayashi will answer this question.
Both the delivery department and the sales and administrative department use AI, with about 90% in the delivery department and 10% in the sales and administrative department. So the AI token fees are already included in our cost of sales this time. Since we are using AI to improve productivity in this area, an increase in token usage will not lower the gross profit. In fact, by leveraging AI, we are aiming to increase the gross profit margin. Therefore, the remaining 10% will be SG&A expenses, but given our company size, this will not have a significant impact. As for the token fees, they are included under system usage fees in our financial results. If by any chance in the future, the amount becomes significant enough to warrant detailed disclosure, we will provide that information.
However, at this stage, the impact is not large enough, so we are not considering disclosing it at this time.
Now the seventh question. In the test market, it seems that the growth rate of the third quarter has slowed to single digits year-over-year. How is the traditional market changing? Is the test market starting to slow down? Kobayashi will also answer this question.
First of all, regarding the test market this time, due to the impact of a major test project in the second quarter, there hasn't been much growth from the second to the third quarter. However, on average, we are seeing steady growth. That said, since we provide a variety of services, from consulting to development and BPaaS, the situation is as described. Regarding this area of testing, we believe that test design and test execution using AI will continue to grow moving forward. As we work to make improvements through consulting, our clients are aiming to reduce costs and also need to speed up development. In these areas, the demand for AI-driven testing is extremely high, and we expect it to continue growing. Our clients also have high expectations for this. So from here on out, we intend to build a solid framework to further expand and strengthen this area.
This is the eighth question. What will the future timing be for recording sales from the fourth quarter and beyond in the sales pipeline? Kobayashi will also answer this question.
Regarding the projects we announced this time, starting from fourth quarter and onward, the figure is about 35%. That's the number for fourth quarter, but after that, it's 50%. Next year with FY 2027 and beyond, and even now, we're working on long-term, large-scale projects for 2028 and 2029, and we're steadily building up our pipeline. Up until now, we didn't have these kinds of large-scale or long-term projects. By building up a long-term pipeline like this, we can steadily lay a solid foundation. Then in the medium term, we fill in with mid-size projects, and in the short term, we maintain our utilization rate with smaller projects. By keeping this kind of balance, we aim to achieve steady growth and earn our customers' trust. This is the sales strategy we are developing.
That concludes the questions we have received. Since we have answered all the questions, we will now conclude the financial results briefing for the third quarter of the fiscal year ending August 2026. Thank you very much for taking the time out of your busy schedules to join us. We hope you will continue to support the SHIFT Group.