Shanghai: the price of intent. How to read $15 billion

On 16 July in Shanghai, Kassym-Jomart Tokayev met Xi Jinping. The announced outcome: more than 70 documents worth over $15 billion. The figure travelled across every wire and became the substance of what was reported about the visit.

I propose taking it apart slowly, because what matters here is not the sum but how it is built.

What “$15 billion” is

First, there is no breakdown. Not one of the seventy-plus documents has a publicly disclosed individual value. Neither a journalist, nor a deputy, nor a citizen can check what the number is made of.

Second, the package mixes things of different natures. Some are agreements in contractual form: Allur with Li Auto on vehicle manufacturing, Astana Group with Chery on a licence, an investment agreement for the first phase of the terminal at Port Kuryk. Some are explicitly labelled memoranda of intent: Samruk-Kazyna with Geely on EV infrastructure. These are fundamentally different commitments, and they sit in one total.

Third, and most telling. Chinese state media do not print the figure at all. Xinhua and Global Times report only that documents were signed “in economy and trade, transport, public finance and the media” — with no number. Euronews, covering the same event, gives $13 billion rather than $15 billion. Nobody reconciled the gap. Before the visit, “around 90 agreements” were trailed; the outcome was “more than 70.”

When one side names a sum, the other declines to confirm it, and Western agencies print a different one, that is not a fact but a claim. Claims can be true. But they must be handled differently from facts.

The format also speaks

This was a working visit — the lowest tier of protocol, neither a state visit nor a summit. The talks were attached to another trip: the president flew to Shanghai for the World Artificial Intelligence Conference, as Xinhua itself notes.

No joint statement was issued. That matters more than it seems. It is precisely joint statements that carry language on sovereignty, on principles, on positions regarding third countries. No document, no trace of what was agreed. The published readouts mention neither Russia, nor Ukraine, nor Taiwan, nor Xinjiang. Absence from a readout does not prove absence from the room, but there is nothing to point to either.

The trade figure needs a footnote

The talks produced the line: 2025 trade turnover of $49 billion, a record. That is Chinese customs data.

Kazakhstan’s own statistics give $34.1 billion for the same year — growth of 13.2%, with China at 23.7% of foreign trade. The fifteen-billion gap is long-standing and publicly unexplained; it is usually attributed to re-export through third countries and to valuation differences.

Both figures can be used. But a publication printing $49 billion without mentioning $34.1 billion is transmitting not a fact but one party’s position.

The number that was not in the reports

Here is an indicator that appeared nowhere in the coverage of the visit.

Kazakhstan’s debt to Chinese lenders: $9 billion in 2022, $9.25 billion in 2023, $9.29 billion at the end of 2024. Three years of near-stasis. And $12.87 billion at the start of 2026 — a rise of nearly 40% in a single year, with over $3.5 billion of new borrowing in 2025.

In May 2026 Kazakhstan placed its first panda bonds — 3.4 billion yuan at 1.9%. And in Shanghai the Development Bank of Kazakhstan signed an agreement with the China Central Depository & Clearing on offshore bond services. That is not a transaction; it is infrastructure — the machinery for further borrowing in yuan.

The debt curve turned upward a year before the visit. For understanding the relationship between the two countries, that is a more substantial fact than any announced investment total.

The historical base you must divide by

Kazakhstan has exactly one publicly auditable list of joint projects with China — the one that triggered the protests of 2019. It was known as the “55 projects.”

By 2022, 25 of 52 planned projects had been realised. Fewer than half. By December 2023 the list had been rewritten: now 45 projects worth $14.5 billion. The count falls, the framing changes, and the totals are not comparable across years.

There are things that were built: the Shymkent refinery upgrade, polypropylene production in Pavlodar region, rapeseed processing in North Kazakhstan. And the reverse: Zhong Mao Group and CITIC Construction both declined to take part in a refinery project; several relocated plants drew objections that their technology had already been retired inside China; a cement plant in Kyzylorda region generated documented complaints about residents’ health and livestock deaths.

The base rate of delivery is under half. The new figure should be discounted against that experience, not against the press release.

I will note separately: Sinopec, Fufeng, Lihua and the auto assembly lines were under construction before the visit. In the coverage they appear as though they were its result.

What is genuinely new

I do not want this to read as denial. There is something new here, and it matters more than the sum.

The organising frame of the relationship is shifting — from oil and rails to compute and connectivity. A strategic partnership between the Ministry of AI and Huawei, procurement of Huawei technology through Samruk-Kazyna, Kazakhtelecom’s agreement with Hengtong on Data Center Valley, a proposed “digital bridge.” This is a different kind of dependency from a pipeline.

Port Kuryk is the most tangible physical deliverable: an investment agreement for the terminal’s first phase, targeting 15 million tonnes of capacity in its first year of operation. Though neither capital expenditure nor a construction start date has been published.

The country’s first pumped-storage hydro plant at Kargaly, around 600 MW, with a contractor from the CCCC group.

And a political detail that passed as courtesy: Xi publicly praised the constitutional reforms and wished success in the coming parliamentary elections. A foreign leader endorsing a domestic order a month before a vote is a statement, not a protocol line.

What was absent

Energy. No oil volumes, no gas, no pipeline agreements, no movement on the nuclear plant with CNNC that was decided back in August 2025.

For a Kazakhstan–China meeting that is a strange silence — particularly after December 2025, when a strike on the CPC terminal led Kazakh crude from Kashagan to be routed to China through Atasu–Alashankou for the first time. Fifty thousand tonnes is a symbolic quantity. Whether that route becomes a structural decision was not discussed publicly.

One more thing was absent: any answer to the China–Kyrgyzstan–Uzbekistan railway, which is being built around Kazakhstan. Construction is under way, with more than five thousand workers and an official completion date of 2030. Transit through Kazakhstan is not logistics; it is the country’s negotiating position. A competing corridor weakens it, and no Kazakh answer was heard in Shanghai.

My view

A multi-vector policy works while there really are several vectors. Once one partner is simultaneously the largest trading partner, a fast-growing creditor, a transit operator and the technology supplier for state data, it turns into rhetoric.

China today accounts for 23.7% of Kazakhstan’s foreign trade and roughly 17–18% of its accumulated direct investment. That is a great deal, but it is not dominance: the Netherlands and the United States are ahead on invested volume. There is still room to manoeuvre.

But I would watch three things rather than the headline sum: how many of the seventy documents become construction sites within twelve months; how fast the yuan-denominated debt grows; and who ends up controlling critical digital infrastructure. Handing Huawei a role in state data is no smaller a sovereignty question than the land that people protested over in 2016. The text of the agreement has not been published and no parliamentary scrutiny has been recorded.

And a final observation I think worth naming. In the first five days after the visit, Kazakhstan’s independent press published not a single analysis of it. Nobody publicly challenged the $15 billion figure. From the outside it is impossible to tell whether that is a shortage of resources or caution. But the result is the same: there was nobody to check.

That, perhaps, matters more than any number in the package.