What the primary evidence shows
None yet, and the reason why is itself part of the finding. No live-web primary research pass has been run on this system — unlike most of this taxonomy’s other systems, everything below comes from a sell-side equity-research triage batch.
That absence of a primary layer is consistent with, not contrary to, the thesis: middleware and cross-embodiment standards are, almost by definition, not owned by any single publicly traded company in a way that generates a dedicated research note to find in the first place.
What sell-side research adds
Essentially no sell-side equity research covers middleware or cross-embodiment standards as a subject, because no publicly traded company’s stock price turns on owning one. The clearest data point on this beat is a market-cap round trip, not a product.
J.P. Morgan’s Korea coverage states a generic preference for “software (robot fleet management, optimisation) over hardware” as the more durable profit pool, then backs that preference with nothing more than a Neutral-rated upgrade of Hyundai Autoever as merely “best-positioned” among a small peer set — no contract, no product, no revenue line attached.
Morgan Stanley’s own Korea IT-services desk is more direct: it calls robot control-software monetisation for Samsung SDS, LG CNS, and Hyundai Autoever “unproven” and “too early-stage to bake into revenue forecasts,” and records that the combined market capitalisation of those three names swung from roughly W35tn to W67tn and back to W42tn within about three weeks on “thematic retail-liquidity volatility,” with no underlying milestone attached to any leg of the move.
The investor read
This is close to a textbook illustration of the taxonomy’s own thesis. The market briefly priced the idea of owning a piece of the robot-software stack and round-tripped nearly the entire move on narrative alone, while the bank covering the theme most closely says on the record that nothing has actually been proven yet.
If P05, Middleware & orchestration monetises at all, the taxonomy’s existing view — that it happens in safety, fleet operations, observability, and support rather than in the middleware itself — has no counter-evidence here and one weak piece of corroboration: nobody, including the analysts paid to find one, has named a company actually capturing that value today.
As with Immune, treat the thin evidence base as an open research task, not a settled conclusion — the productive next step is primary research on standards-body activity and commercial-support vendors, not another sell-side batch.
一手证据显示什么
目前尚无一手证据，而其原因本身就是发现的一部分。这一系统尚未经历一手网络调研——与本分类法其他大多数系统不同，下文全部内容都来自一批卖方股票研究的筛选整理。
而这种一手研究层的缺失，恰恰与论点本身相符，而非相悖：中间件与跨本体标准，几乎从定义上就不属于任何一家上市公司，因而一开始就不会催生一份专门的研究报告去发现它们。
卖方研究补充了什么
几乎没有卖方股票研究把中间件或跨本体标准作为一个独立主题来覆盖，因为没有任何一家上市公司的股价，是靠拥有某个中间件来支撑的。这条赛道最清晰的数据点，是一次市值的往返涨跌，而非一款产品。
摩根大通的韩国研究团队泛泛地表示，相较硬件，“软件（机器人车队管理、优化）”是更持久的利润池，但支撑这一偏好的，仅仅是把现代 Autoever 上调至“中性”评级、称其为一小群同业中“定位最佳”的公司——没有具体合同，没有具体产品，也没有任何收入线。
摩根士丹利自己的韩国 IT 服务研究团队则说得更直白：其认为三星 SDS、LG CNS 与现代 Autoever 在机器人控制软件方面的变现能力“尚未得到验证”，“为时过早，不足以纳入营收预测”；该团队还记录到，这三家公司的合计市值在约三周内从约 35 万亿韩元飙升至约 67 万亿韩元，随后又回落至约 42 万亿韩元，原因是“主题性散户流动性波动”，而整个涨跌过程中并没有任何实质性里程碑事件作为支撑。
给投资者的启示
这几乎是本分类法自身论点的一个教科书式例证。市场曾短暂地为“拥有机器人软件栈的一部分”这一构想定价，随后几乎完全靠叙事本身把整段涨幅又跌了回去；而对这一主题跟踪最紧密的银行，白纸黑字地表示目前尚无任何东西被真正证实。
如果说 P05，中间件与编排 终究会变现，本分类法既有的看法——变现会发生在安全、车队运营、可观测性与支持环节，而非中间件本身——在此并未遇到反证，只得到了一条薄弱的佐证：包括受雇专门去寻找答案的分析师在内，没有任何人能指出今天究竟是哪家公司在真正捕获这部分价值。
与“免疫”系统一样，应把这薄弱的证据基础当作一项有待推进的研究任务，而非已经定型的结论——真正有价值的下一步，是针对标准组织活动与商业支持供应商的一手研究，而不是再做一轮卖方筛选。