Is ExaWizards profitable in 2026?

Yes. For FY2026/3 (ended March 31, 2026), ExaWizards reported consolidated net profit of Y1,513 million, swinging from a Y2,576 million net loss the prior year, on net sales of Y11,996 million (+22.3% YoY), per its May 12, 2026 TDnet filing.

Who owns ExaWizards now?

Sumitomo Mitsui Financial Group (SMFG) became ExaWizards' largest shareholder effective April 16, 2026, after subscribing to 9,550,000 new shares (about Y5.4 billion, ~$33.8M) for a 10% voting stake announced March 31, 2026, alongside a financial-sector AI alliance.

Could AI replace ExaWizards' consulting business?

Partially already happening: ExaWizards' AI Solution Services (custom-build consulting) segment revenue declined in FY2026/3, which the company attributed to fewer bespoke build projects as generic generative AI tools reduce demand for custom development, per its February 12, 2026 Q3 results presentation.

Who is ExaWizards' biggest competitive threat?

Domestic Japanese system integrators NTT Data, Fujitsu, NEC, and Hitachi, which already have deeper government and enterprise relationships and face no structural bar to shipping rival gen-AI products; Fujitsu already ships its own Takane LLM and NTT its tsuzumi 2 LLM (October 2025).

Does ExaWizards build its own AI models?

No. ExaWizards markets 'rapid integration of the latest generative AI models' into its government products, per a February 17, 2026 press release, meaning its exaBase platform orchestrates third-party frontier models (OpenAI, Google, Anthropic-class) rather than owning a proprietary foundation model.

How widely is ExaWizards used by Japanese governments?

As of February 2026, ExaWizards' exaBase Generative AI for Governments was used by 25 of Japan's 47 prefectural governments (53%) and over 200 government agencies and municipalities, distributed through a network of 60+ certified reseller partners nationwide, per company investor relations data.

AI companies and labs · Japan

ExaWizards

exawizards.com ↗

Japanese firm that wires government agencies and banks into outside AI models, not its own.

Losing its protection

Assessed 26 July 2026 · Evidence as of 26 July 2026

The structural protection is eroding.

On the recordSince 26 July 2026 · Checkable from August 2026 · Every call on the record Checkpoint: Q1 FY2027/3 earnings, the first test of whether AI Products growth holds 1.5 to 1.7x or slows toward integrator pace

The call

ExaWizards turned profitable on government and bank distribution it built early, but it owns no model of its own and faces bigger integrators who could copy that playbook.

ExaWizards sells artificial intelligence software to Japanese government agencies and banks, using AI models from companies like OpenAI and Google rather than building its own. The company is hard to replace right now because it has deals with 25 prefectures and 200+ government offices, plus a major bank putting money in, but this advantage is fragile: bigger Japanese tech companies like NTT Data and Fujitsu already sell software to those same government customers and could easily add AI products to their existing offers, while smaller government offices could just use OpenAI or Google's tools directly through Microsoft Azure. The picture could change if SMFG bank's AI tool performs well and locks in long-term work that makes ExaWizards essential, or if the big tech companies stay out of the government AI business for the next year or two.

Why

AI pressure: Generic frontier models (GPT, Gemini, Claude) used directly by enterprises. Clients now build custom AI straight on frontier models, and ExaWizards' consulting segment is already shrinking because of it. Could AI do the core job? Yes, likely.

Bigger players: NTT Data, Fujitsu, NEC, Hitachi (Japanese system integrators with existing government ties). Fujitsu already runs its own Takane LLM and NTT Data has tsuzumi; both could push direct gen-AI products into ExaWizards' prefecture accounts. Could a giant copy it? Yes, easily.

What protects it. A two-year head start selling into 25 prefectures, 200+ agencies and 60+ resellers, plus a fresh SMFG capital tie and engineer secondment for CFO Agent.

What would change this call

Fujitsu or NTT Data launch a direct generative-AI government product within 12-18 months. Would show ExaWizards' distribution head-start into prefectures is being replicated by incumbent SIs.
AI Products segment growth rate in FY2027/3 Q1 results (expected August 2026). Compression toward incumbent SI growth pace would signal the distribution moat is eroding.
SMFG's CFO Agent rollout expands to other Japanese megabanks or stalls within ExaWizards. Expansion would show data lock-in hardening; stalling would show the alliance is shallow.

Key numbers

FY2026/3 net sales Â¥11,996M, up 22.3% YoY (2026-03-31)
AI Products segment growing 1.5 to 1.7x YoY (2025-11-11)
SMFG bought 10% stake for about $33.8M (2026-03-31)

Named risks

AI Solution Services consulting revenue declining as bespoke build work is commoditized by generic gen-AI tools (per Feb 2026 results). Severity: high
exaBase wraps third-party frontier models with no proprietary foundation model, limiting structural differentiation. Severity: high
Domestic system integrators (NTT Data, Fujitsu, NEC, Hitachi) face no structural bar to shipping competing gov AI products within 12-18 months. Severity: high
Stock highly volatile with 52-week range roughly Y227 to Y1,445, tied to AI-sector sentiment. Severity: medium
New subsidiary launches (Kyushu, Interaction Design Lab) add near-term cost before proven revenue return. Severity: low

Rivals, as read in July 2026

Rival	Direction	Evidence
NTT Data	contesting	Ships own tsuzumi 2 Japanese LLM (Oct 2025); holds deeper govt relationships, untested vs ExaWizards
Fujitsu	contesting	Already ships proprietary Takane LLM, could launch direct gen-AI government products
NEC	contesting	Deeper enterprise/government relationships than ExaWizards, no structural bar to competing
Hitachi	contesting	Existing government/enterprise accounts could host rival gen-AI offerings, untested as of mid-2026
OpenAI, Google, Anthropic, Microsoft	contesting	Frontier model access partly blocked by ExaWizards' local distribution but reachable via Azure-through-Japanese-partner routes

Questions people ask

Is ExaWizards profitable in 2026?: Yes. For FY2026/3 (ended March 31, 2026), ExaWizards reported consolidated net profit of Y1,513 million, swinging from a Y2,576 million net loss the prior year, on net sales of Y11,996 million (+22.3% YoY), per its May 12, 2026 TDnet filing.
Who owns ExaWizards now?: Sumitomo Mitsui Financial Group (SMFG) became ExaWizards' largest shareholder effective April 16, 2026, after subscribing to 9,550,000 new shares (about Y5.4 billion, ~$33.8M) for a 10% voting stake announced March 31, 2026, alongside a financial-sector AI alliance.
Could AI replace ExaWizards' consulting business?: Partially already happening: ExaWizards' AI Solution Services (custom-build consulting) segment revenue declined in FY2026/3, which the company attributed to fewer bespoke build projects as generic generative AI tools reduce demand for custom development, per its February 12, 2026 Q3 results presentation.
Who is ExaWizards' biggest competitive threat?: Domestic Japanese system integrators NTT Data, Fujitsu, NEC, and Hitachi, which already have deeper government and enterprise relationships and face no structural bar to shipping rival gen-AI products; Fujitsu already ships its own Takane LLM and NTT its tsuzumi 2 LLM (October 2025).
Does ExaWizards build its own AI models?: No. ExaWizards markets 'rapid integration of the latest generative AI models' into its government products, per a February 17, 2026 press release, meaning its exaBase platform orchestrates third-party frontier models (OpenAI, Google, Anthropic-class) rather than owning a proprietary foundation model.
How widely is ExaWizards used by Japanese governments?: As of February 2026, ExaWizards' exaBase Generative AI for Governments was used by 25 of Japan's 47 prefectural governments (53%) and over 200 government agencies and municipalities, distributed through a network of 60+ certified reseller partners nationwide, per company investor relations data.

Timeline

2026-06-22ExaForward Kyushu subsidiary approved; business start planned November 2026
2026-06-09Exa Interaction Design Lab subsidiary approved for AI agents and local LLM R&D
2026-05-12FY2026/3 results: net profit Y1,513M, reversing prior year Y2,576M loss
2026-05-12Exa HomeCare merged into ExaMD to consolidate CareWiz nursing-care line
2026-04-16SMFG becomes largest shareholder with 10% voting stake
2026-03-31SMFG capital and business alliance announced, ~$33.8M share subscription
2026-02-17Press release touts rapid integration of latest generative AI models into gov products
2026-02-12Q3 FY2026/3 results show AI Solution Services segment revenue decline
2026-01-30Business alliance with Dirbato for AI-agent business process consulting
2025-09-30exaBase Empower HR analytics product launched with free trials

Revision history

26 July 2026

First assessed. Losing its protection. The text above is fixed as of this date and is not edited afterwards.

In the same band

Safe Superintelligence Inc.Mistral AI Cohere