Michibiki No. 7 Reached Orbit. The Constellation Is Still Five Years Away

Hello, this is Ryuta Hamamoto from TIMEWELL.

Early on 11 August 2026, an H3 rocket lifted off from Tanegashima and placed the quasi-zenith satellite Michibiki No. 7 into its planned orbit. The coverage ran as "Japan's own GPS launched successfully."

It would be a waste to file this under "nice rocket." Positioning is the kind of infrastructure that sits underneath everything and gets no thought at all until it stops, at which point you discover how much was resting on it.

That said, some of what is being said about this launch is not factually right. Having read the Cabinet Office material, I want to set out both the good news and the part that is not finished. Skipping the second would make the article pointless.

What actually happened on 11 August

Facts first. At 04:23:31 on 11 August 2026, H3 rocket Flight 9 lifted off from the Tanegashima Space Center and, about thirty minutes later, separated Michibiki No. 7 and placed it in its planned orbit¹. No. 7 occupies a quasi-geostationary orbit at 185 degrees east².

The Quasi-Zenith Satellite System, Michibiki, is a satellite positioning system that Japan manages and operates. It flies at roughly 36,000 kilometres, about the altitude of geostationary orbit. The "quasi-zenith" name comes from an orbit designed so that one satellite is always near the zenith over Japan².

The mechanism is simple. Satellites broadcast one-way to receivers on the ground, and signals from four or more satellites fix a position and a time². That "four or more" is the hinge of everything that follows.

Why seven satellites is the goal

This is the substance.

Michibiki began four-satellite operation in 2018 and has been running with five since July 2025, when No. 6 joined². But the way it is used today is that it broadcasts GPS-compatible signals and is used together with GPS². Michibiki is not replacing GPS; it is sharpening it.

So what would change that? Seven satellites.

At seven, four or more Michibiki satellites are visible above Japan at all times. Since positioning needs four, Michibiki alone becomes sufficient. The term for this is sustained positioning².

The Cabinet Office material states the benefit of the seven-satellite constellation in these words:

Realises social infrastructure that does not depend on other countries' GNSS even in the event of failure (also important from a security standpoint)²

The government itself frames this in security terms.

To be clear about what this is not. This is not an argument that some country is going to switch GPS off. I have no material for that claim. The problem is structural: having no alternative is itself what removes your options. Whatever anyone's intent, the party that depends on a system has nothing to fall back on the moment that system becomes unavailable — through a fault, a disaster, or interference alike.

Beyond seven, the Cabinet Office is developing toward eleven satellites, at which point positioning survives the failure of any single satellite². Seven is described as the bare minimum.

If you want the wider economic security picture first, the expert panel on economic security sets it out from primary sources.

It is not seven yet. FY2031 is the target.

This is what gets dropped, in coverage and in internal briefings alike.

This success takes Michibiki to six satellites. Not seven.

The reason is stated plainly in the Cabinet Office material:

No. 5 was lost in December 2025 in the launch failure of H3 rocket Flight 8²

On 22 December 2025, the second burn of the H3 Flight 8 second-stage engine failed to start properly and shut down early, leaving Michibiki No. 5 short of its planned orbit³. The original plan had been to place No. 5 in quasi-zenith orbit, No. 6 in geostationary orbit and No. 7 in quasi-geostationary orbit, completing seven satellites within FY2025³. One of those is gone.

Under the current plan, seven satellites arrives in FY2031, when a successor to No. 3 and a new No. 8 are launched, with No. 8 carrying capability equivalent to or better than No. 5³.

Which means sustained positioning is roughly five years out from here.

I am not writing this to spoil the mood. The opposite. Read as "we are independent now," the preparation that those five years require stops happening. Michibiki today is designed to be used with GPS, and Michibiki alone does not fix a position where GPS is unavailable. For anyone writing a continuity plan, that premise has not yet changed.

None of which makes this launch small. After the Flight 8 failure came root cause analysis and an assessment of the effect on subsequent vehicles, and then Flight 9 flew and worked. Flying again after a failure is considerably harder than an unbroken run of successes, in my view.

What six centimetres actually changes

Now accuracy. Getting the numbers right makes this concrete.

Cabinet Office material puts typical GPS positioning error at 5 to 10 metres². Michibiki broadcasts augmentation signals that correct this, in three main services².

Service	Error	Coverage
SLAS (sub-metre)	under 1 metre	Japan
CLAS (centimetre)	6 cm or better	Japan only
MADOCA-PPP	decimetre-level	Asia and Oceania

CLAS at 6 centimetres or better is what people mean by "a few centimetres." With a condition attached: Japan only. CLAS relies on domestic electronic reference station data, so it does not work outside the country². Overseas, decimetre-level MADOCA-PPP has been available since FY2024².

What does six centimetres change? The Cabinet Office example is specific enough to be useful. In a smart agriculture application in Hokkaido, the track of a tractor working a field changed like this²:

• Metre-level (until 2023): "accuracy is ten times better than GPS, but still not practical"
• 6 centimetres (2024 onward): "at centimetre level it finally becomes practical"

"Ten times better than GPS and still not enough" is the line that landed for me. Run a tractor automatically and being a metre off means driving over the neighbouring row. Ten-metre accuracy and centimetre accuracy are not different degrees of the same thing; they enable different things.

The same holds for autonomous driving, drone inspection of facilities, automated snow clearance and container tracking in logistics².

This also works as a return-on-investment argument. Labour shortage is coming, reliably. Automating core industries requires, underneath it, centimetre-level positioning available anywhere as a matter of course. Launches look expensive, but this is base infrastructure of the same class as roads and electricity. It is not something individual companies can provision for themselves.

Distributing it across Asia and Oceania

Michibiki's signals do not stop at Japan. The distribution area is Asia and Oceania².

What matters here is the non-positioning services. The Early Warning Satellite Service (EWSS) broadcasts earthquake and tsunami information issued by disaster agencies such as the Japan Meteorological Agency, and it can be distributed to overseas areas across Asia and Oceania². The Cabinet Office material cites delivery in Fiji, in the South Pacific, of disaster information including earthquakes and tsunamis as well as material produced by local disaster agencies².

That is understated but consequential. In a disaster that takes down terrestrial networks, information falling from above still arrives. Offering that to neighbouring countries means carrying part of their disaster infrastructure.

MADOCA-PPP has the same shape: decimetre-level augmentation offered to countries across Southeast Asia, Australia and the Pacific islands, with operators working offshore away from the coast, such as fisheries, among the anticipated users².

How to read this diplomatically and economically depends on where you sit. My own phrasing would be less "seizing the initiative" and more becoming the option that gets chosen. Positioning infrastructure is not easily swapped once it is embedded. If another country's industry and disaster response come to sit on top of it, the relationship persists on its own. It is not something you obtain by pushing; the party that put it there first tends to end up chosen.

The dependence is not only positional

Here is what I most wanted to write.

Positional dependence is one instance of a shape that repeats. Some numbers, all from government publications.

Energy self-sufficiency is 15.3% (FY2023), the lowest among the G7⁴. Imported fossil energy accounts for over 80% of primary energy supply, also high by G7 standards⁴.

Food self-sufficiency on a calorie basis is 37% (FY2025), down one point from the previous year⁵.

And several dependencies rarely expressed as numbers at all.

Communications. When terrestrial networks fail, Japan has no domestic means of bringing communications down from above. Direct-to-device satellite connectivity is led by overseas operators, and Japanese regulation has yet to catch up on unlicensed bands. I covered that in satellites connecting directly to devices and the FCC proposal. There is an asymmetry worth noticing: positioning is being brought in-house, communications is not.

AI models. Foundation model supply is concentrated in US and Chinese firms. The more deeply they are embedded in operations, the larger the impact when one becomes unavailable.

Components and hardware. Network equipment and solar panels show pronounced concentration. The IEA reports that China holds over 80% of global manufacturing capacity across the main solar panel stages — polysilicon, ingots, wafers, cells and modules⁶.

To be explicit. This is not an argument that buying from a particular country is wrong. Specialisation has economic logic and is, on its own terms, a sound decision. The problem is the absence of an alternative route, and not knowing you depend on something. The first is a question of options, the second of awareness, and the second is the more troublesome of the two.

As a concrete case, a component inside a component was communicating externally without the manufacturer of the finished product knowing — I covered that in a Royal Navy sea drone calling China. That is a military example, but the structure is identical in a factory security camera.

If you want to map which countries' regimes your procurement and technology management currently touch, the free export control readiness check gives you a starting position.

Translating "sovereignty" into operational language

To recap.

On 11 August 2026, H3 Flight 9 placed Michibiki No. 7 into its planned orbit. That takes Michibiki to six satellites. Because No. 5 was lost in the December 2025 failure of H3 Flight 8, the target for seven is FY2031. Sustained positioning, using Michibiki alone, is about five years out.

On accuracy: against 5 to 10 metres for GPS, CLAS delivers 6 centimetres or better — within Japan only, with decimetre-level MADOCA-PPP across Asia and Oceania.

And the dependence extends well past positioning. Energy 15.3%, food 37%, no domestic satellite communications network, AI models and key components concentrated in particular countries.

"Restoring sovereignty" is a phrase large enough to be hard to act on. I translate it as: if it stops, is there an alternative?

That version is company-sized work. List the paths by which your operations stop, and check each for an alternative. Positioning, communications, power, materials, cloud, AI models. Tabulate the ones with no alternative, then set the cost of building one against the loss from a stoppage. It looks like a national question and is in fact only ever solved one company at a time.

Honestly, the thing that caught me while writing this was that FY2031 was further out than I had assumed. I had read it as "independent now" myself at first. The same applies to statutes and budgets: the better the news, the more worth checking what conditions remain inside it. "Achieved" and "on the way to being achieved" are different things, and the second leaves the question of how you get through the interval.

Designing that interval is not solely a matter for government. Very few companies could currently tell you what stops if positioning does. That seems like the place to start.

If you want to review a management framework spanning several countries' regimes, or the dependency structure of your supply chain, TRAFEED may be a useful reference point, and you can bring your own situation to us here.

---

Footnotes

1. H3 rocket Flight 9 lifted off from the Tanegashima Space Center at 04:23:31 on 11 August 2026 and, approximately thirty minutes later, separated the quasi-zenith satellite Michibiki No. 7 and placed it in its planned orbit. https://sorae.info/ssn/20260811-h3f9.html — launch time and window per the Cabinet Office Michibiki official site. https://qzss.go.jp/info/information/qzs-7_260810-2.html ↩

2. Cabinet Office, National Space Policy Secretariat, "Toward the launch of Quasi-Zenith Satellite System Michibiki No. 7 (latest developments)" (August 2026, pre-launch briefing material for H3 rocket Flight 9). Michibiki's altitude of approximately 36,000 km (about that of geostationary orbit); positioning requiring signals from four or more satellites to determine position and time; four-satellite operation from 2018 and five-satellite operation from July 2025 with the addition of No. 6; use together with GPS via GPS-compatible signals; typical GPS positioning error of 5 to 10 metres; the SLAS (sub-metre, under 1 m), CLAS (centimetre, 6 cm or better, limited to Japan as it requires domestic electronic reference station data) and MADOCA-PPP (decimetre-level, Asia and Oceania, service from FY2024) services; the statement that "No. 5 was lost in December 2025 in the launch failure of H3 rocket Flight 8"; the seven-satellite constellation realising "sustained positioning" in which positioning is possible using Michibiki alone; the stated benefit of "social infrastructure that does not depend on other countries' GNSS even in the event of failure (also important from a security standpoint)"; development toward an eleven-satellite constellation; No. 7's placement at 185 degrees east; the Hokkaido smart agriculture example in which metre-level accuracy is described as "ten times better than GPS, but still not practical" and 6 cm as "at centimetre level it finally becomes practical"; and the Early Warning Satellite Service (EWSS) being distributable to overseas areas across Asia and Oceania with delivery in Fiji — are all from this material. https://www.jaxa.jp/projects/files/youtube/h3f9/jaxa_doc02_20260805.pdf ↩ ↩² ↩³ ↩⁴ ↩⁵ ↩⁶ ↩⁷ ↩⁸ ↩⁹ ↩¹⁰ ↩¹¹ ↩¹² ↩¹³ ↩¹⁴ ↩¹⁵ ↩¹⁶ ↩¹⁷ ↩¹⁸ ↩¹⁹

3. In the launch of H3 rocket Flight 8 on 22 December 2025, the second burn of the second-stage engine failed to start properly and shut down early, so Michibiki No. 5 could not be placed in its planned orbit. The original plan was to launch No. 5 into quasi-zenith orbit, No. 6 into geostationary orbit and No. 7 into quasi-geostationary orbit, completing the seven-satellite constellation within FY2025. Under the current plan, a successor to No. 3 and a new No. 8 launch in FY2031, with No. 8 carrying capability equivalent to or better than No. 5, completing the seven-satellite constellation. https://sorabatake.jp/44151/ — on the Flight 8 launch failure, statement by the Minister of Education, Culture, Sports, Science and Technology https://www.mext.go.jp/b_menu/daijin/detail/mext_00651.html ↩ ↩² ↩³

4. Agency for Natural Resources and Energy, "Energy White Paper 2025." Per the text, "Japan's energy self-sufficiency was 15.3% as at FY2023, the lowest level among the G7 countries," and imported fossil energy accounts for over 80% of primary energy supply, a high level compared with other G7 countries. https://www.enecho.meti.go.jp/about/whitepaper/2025/html/1-2-1.html ↩ ↩²

5. Ministry of Agriculture, Forestry and Fisheries, "Japan's food self-sufficiency rate." For FY2025 the calorie-basis food self-sufficiency rate was 37%, down one point from the previous year owing to a decline in domestically supplied calories. https://www.maff.go.jp/j/zyukyu/zikyu_ritu/012.html ↩

6. According to International Energy Agency reporting, China accounts for over 80% of global manufacturing capacity for the main solar panel elements (polysilicon, ingots, wafers, cells and panels/modules). https://www.jetro.go.jp/biznews/2022/07/d345fae73fe8499c.html ↩