Agilink

1.1 Identity, positioning, and disclosed operating scale

Agilink's current official positioning is tightly focused on dexterous manipulation systems rather than general-purpose humanoids: the company says it designs and manufactures high-DoF dexterous hands and adaptive grippers for research, industrial automation, and service robotics. The current product lineup on official surfaces centers on the OmniHand 3 family and OmniPicker 3, while the public product catalog emphasizes application fit, mounting, power, communication, and control-software integration. On scale, the strongest currently visible numbers remain company claims rather than independently benchmarked disclosures: Agilink's homepage advertises No.1 market share in Q1 2026, 20,000+ products shipped, and 50,000+ hours of real-world data. Those claims point to real commercial and deployment ambition, but they do not yet substitute for verified revenue, customer concentration, or repeat-order disclosure.[CO001, CO002, CO006, CO007, CO016]

1.2 Leadership, governance, and parent dependence

The clearest founder signal in the assigned pool is Xiong Kun, whom both Baidu Baike and Sina associate with the Agilink spinout and with prior Tencent Robotics X and robotics engineering experience. That gives the company credible technical founder-market fit for dexterous end-effectors, but it also highlights key-person dependence because the public record is much richer on the founder than on broader management or board oversight. Governance visibility remains imperfect. Baidu Baike names Wang Chuang as legal representative, while Sina's May 2026 article names Deng Taihua, creating a live discrepancy in basic corporate-control data. The official about page further confirms that Agilink originated inside AgiBot before becoming independent in January 2026, so diligence still needs to test how much product roadmap, sales mix, IP, and transfer-pricing exposure remains tied to the parent ecosystem.[CO003, CO005, CO011, CO012, CO013]

1.3 Funding pace, stakeholders, and disclosure gaps

Public evidence supports a very fast capital-formation story. The about page says Agilink became independent in January 2026, while Baidu Baike and Sina describe multiple rounds completed within months and a unicorn-level valuation by May or June 2026. Baidu Baike also points to a parent-controlled cap table, saying Shanghai Zhiyuan New Innovation Technology holds roughly 67% of shares and that Tencent- and SAIC-linked vehicles became shareholders. That combination matters for valuation and control analysis: Agilink appears to have attracted strong investor pull, but public sources still do not expose board composition, investor rights, liquidation preferences, or related-party commercial terms with AgiBot. The chapter can therefore establish stakeholder shape, but not yet clean governance economics.[CO008, CO009, CO010, CO013, CO016]

1.4 Milestones and external risk context

The dated milestone trail is unusually dense for such a young standalone company: official and third-party sources support a path from internal incubation in 2023 to launch milestones in 2025, independence in January 2026, new Omni products in April, unicorn valuation markers by May and June, and an October 2026 dexterity demonstration that helped keep the story visible. That said, the sector backdrop is not uniformly favorable. Humanoids Daily reports that Beijing warned China's humanoid and embodied robotics sector against "highly repetitive products" and bubble dynamics. For Agilink, that means strong early momentum is arriving in a market that may reward differentiated execution but punish copycat hardware and overcapitalized supply chains.[CO003, CO008, CO009, CO014, CO015]

1.5 Exhibits

2.1 Market boundary and buyer map

The assigned pool points to a narrower market than generic humanoid-robot headlines imply. Agilink's official positioning is centered on dexterous hands and grippers, while the AGIBOT store frames selection around the object, movement, and host robot instead of the whole robot stack. That keeps the included spend at the end-effector, integration, and manipulation-data layer and excludes broader humanoid locomotion, full robot platforms, and general-purpose foundation-model budgets that the evidence does not isolate. Buyer and payer evidence also clusters around three initial groups: research teams collecting manipulation data, robot OEMs or integrators fitting hands into humanoid and general-robot platforms, and industrial automation teams using grippers where the workflow favors reliability over anthropomorphic dexterity. [CM001, CM002, CM003, CM004, CM005, CM009, CM012]

2.2 Sizing lenses and demand drivers

The evidence supports multiple sizing lenses, but not a single clean TAM. IFR's 2026 release places China at roughly 2 million operational industrial robots and 54% of annual global industrial-robot installations, which is the broadest credible demand context in the assigned pool. The same source shows domestic suppliers increasing share from 30% of Chinese installations in 2020 to 57% in 2024, which matters because Agilink is selling Chinese-designed end-effectors into a policy-supported localization cycle. Humanoids Daily adds a second lens by describing more than 50 embodied-intelligence financings and roughly RMB 20 billion of Q1 2026 capital in China, while Agilink-specific shipment figures from Humanoids Daily and Baidu offer a third lens on current throughput. Together those inputs justify a demand narrative, but they remain installed-base, capital, and unit-shipment proxies rather than a revenue-backed market stack. [CM006, CM007, CM008, CM009, CM010, CM012, CM014]

2.3 Adoption constraints and unresolved market proof

The strongest disconfirming evidence is not that demand disappears, but that broad humanoid commercialization remains slower and narrower than unicorn narratives imply. IFR explicitly says universal humanoid factory or household adoption will not happen in the near- or medium-term future and instead expects AI adoption with traditional industrial robotics over the next five to ten years. That pushes Agilink's nearer commercial opportunity toward concrete, integration-ready workflows rather than a generic claim that the total humanoid market is already open. The assigned pool also leaves two valuation-relevant gaps unresolved: there is no independent denominator for Agilink's claimed Q1 2026 China market-share leadership, and there is no public revenue, ASP, or external-customer mix disclosure that would translate the shipment narrative into defensible TAM, SAM, or SOM ranges. [CM003, CM011, CM013, CM014, CM015]

2.4 Exhibits

3.1 Competitive set is broader than hand-vs-hand

The retained pool shows that Agilink competes on more than a single product axis. Its own official surfaces split the market between dexterous hands for multi-finger manipulation and grippers for simpler object-handling work, which means the practical substitute is often not another five-finger hand but a lower-complexity gripper or other task-specific tooling. The same official guidance also says buyer fit depends on mount, payload, power, communication, and software interfaces, so competition is partly a component-selection process inside a broader robot program, not just a catalog comparison between end effectors.[CP001, CP002, CP012]

3.2 Direct peers, adjacent vendors, and incumbents attack different buyer priorities

Agilink's clearest adjacent pressure in the retained 2026 pool comes from ROBOTERA, which combines a dexterous-hand architecture with a broader robotics platform and a live commercialization story in logistics. Shadow Robot represents a different incumbent threat: less evidence of recent scaled deployment in this pool, but long-standing research credibility and a 24-degree-of-freedom hand that continues to anchor the high-end benchmark set. Agilink's own position sits between those poles: newer and faster-moving than Shadow Robot, but still proving whether its early shipment and market-share claims translate into durable, independently verified share against other ambitious embodied-AI vendors.[CP003, CP004, CP005, CP006, CP007, CP012, CP013, CP014]

3.3 Switching costs exist, but public pricing and win-rate visibility are weak

The most visible switching-cost levers are integration and developer workflow rather than headline price. Agilink's retained sources emphasize mechanical fit, communication protocols, and software access, while Unite.ai's recap of published specifications points to SDKs, simulation assets, and ROS 2 reference nodes that can create meaningful engineering stickiness once a hand is integrated. Even so, the retained sources do not disclose public pricing for Agilink, ROBOTERA, or Shadow Robot, so buyers and investors cannot benchmark feature-price tradeoffs cleanly from public evidence. That opacity also makes it harder to know whether Agilink wins on capability, discounting, parent-network pull, or bundled support.[CP002, CP009, CP010, CP015]

3.4 Moat durability still depends on proving independence from the parent ecosystem

The retained evidence suggests Agilink's current edge is partly channel and ecosystem speed. Official and affiliate sources say the company was formed inside AGIBOT, Baidu records AGIBOT's Shanghai Zhiyuan affiliate with roughly two-thirds ownership, and C Capital ties the hand business to AGIBOT robots already working on customer factory floors. That can accelerate deployment, but it also means much of the visible proof sits inside a company-controlled or financially aligned network. ROBOTERA already markets its own direct-drive hand and in-house stack with multi-site deployments, while Shadow Robot remains a recognizable research benchmark, so Agilink still needs more independent win/loss, reliability, and pricing proof to show that its advantage is durable rather than a fast-moving, parent-enabled first wave.[CP003, CP008, CP011, CP013, CP014, CP016]

4.1 Revenue model and monetization

Agilink's public commercial surface still looks like an industrial hardware supplier, not a disclosed recurring-software business. The AGIBOT store only presents OmniHand dexterous hands and OmniPicker grippers, while the corporate site describes dexterous manipulation systems and end-effector solutions rather than subscriptions or usage-based software. The sales path is also consultative: buyers are told to choose around their application, then confirm mounting, power, communication, software, budget range, and timeline with sales. That implies revenue likely starts with hardware orders plus potentially bundled integration support, but public materials do not disclose list prices, realized ASPs, contract length, discounts, or any recognized-services revenue.[CI002, CI003, CI009]

4.2 Traction and unit-economics proxies

Public traction signals are stronger than public economics. Company-issued September 2026 coverage says Agilink has shipped more than 15,000 dexterous hands, while Baidu Baike's June 2026 summary says cumulative OmniHand deliveries exceeded 8,000 units and gripper shipments exceeded 10,000 units. The official about page also shows a rapid operating ramp: Agilink says it has 60+ team members today, is scaling toward 100+ by Q4 2026, launched OmniHand and OmniHand Pro in August 2025, hit its 1,000th hand in December 2025, and unveiled multiple Omni 3 products in 2026. Those signals support real production activity, but they still do not convert shipment counts into revenue, gross profit, or customer quality because the company does not break out realized pricing, external-customer mix, or service costs.[CI004, CI005, CI006, CI007, CI008, CI014]

4.3 Capital adequacy and financing dependence

Agilink has clearly been able to raise capital, but current adequacy is not publicly underwritable. Crunchbase News reports that the company reached a $1 billion valuation after four funding rounds since its January 2026 spinout and that the new capital would fund model development, data, and hardware. Agilink's own about page also says a June 2026 round pushed post-money valuation above $1 billion. That financing momentum matters because the company is adding headcount and shipping increasingly sophisticated hardware, both of which usually pull forward working-capital needs, inventory commitments, warranty reserves, and support costs. Baidu Baike's ownership summary also says the Agibot parent still holds roughly 67% of the shares, so intercompany dependence remains a real consideration even as Agilink broadens into the external robotics market.[CI004, CI010, CI011, CI012]

4.4 Financial verdict and diligence blockers

The financial verdict is therefore mixed: commercialization appears real, but revenue quality and capital sufficiency are not yet transparent enough for a clean underwriting case. Shipment and fundraising milestones show momentum, yet none of the reviewed public sources disclose revenue, gross margin, burn, runway, debt obligations, customer concentration, or the split between Agibot-internal demand and third-party buyers. Humanoids Daily's reporting on Beijing's warning about sector "bubbles" and repetitive humanoid products adds a credible adverse backdrop because it raises the bar on differentiation and capital discipline exactly when Agilink is scaling production. Until the company provides SKU-level ASPs, product-family margins, liquidity, and parent-transaction detail, the chapter supports a research-more stance rather than a conviction view on margin path or financing resilience.[CI007, CI010, CI012, CI013, CI014]

4.5 Exhibits

5.1 Product scope is clear, but the offer is still catalog-led

Agilink's official surfaces consistently define the company around dexterous manipulation rather than a broad robot platform: the about page frames the company as a builder of robotic hands and grippers, and the AGIBOT store asks buyers to choose an end effector around the object, movement, and host robot in the target application. That positioning is useful because it anchors the product in a customer workflow instead of a generalized embodied-AI narrative. The public catalog is still narrow, however. On the main dexterous-hand collection page, the visible current lineup is two OmniHand variants plus OmniPicker 3, which suggests a focused end-effector business rather than a fully modular stack with many production SKUs. Baidu Baike's scenario summary broadens the same story from catalog to go-to-market: data-collection use cases, robot-integration programs, and industrial or research deployments. [CE001, CE002, CE003, CE008]

5.2 Publicly evidenced architecture centers on end-effector mechanics, tactile sensing, and integration dependencies

The strongest product proof in the assigned pool is the flexible-material demo record. Two independent write-ups say OmniHand 3 Ultra handled silk-thread separation, needle threading, hooping, and stitching, and both describe a 21-active-DOF, direct-drive hand with tactile sensing. The Rundown adds a training path that starts with teleoperated data collection and then reinforcement-learning refinement, which is more concrete than generic “AI-powered” positioning. At the same time, the deployment burden is still pushed to integrators: the store tells buyers to validate mount, power, communications, and control-software compatibility before purchase. That matters because Humanoids Daily places the OmniHand 3 Ultra-T inside AGIBOT's broader “One Body, Three Intelligences” stack, so the public record points to meaningful ecosystem dependence rather than a separately documented Agilink software/control platform. [CE005, CE006, CE007, CE009, CE010, CE011, CE014]

5.3 Release cadence and shipment claims support momentum, but not yet underwriting-grade maturity

Agilink's official timeline shows an unusually fast 2025-2026 sequence: OmniHand and OmniHand Pro in August 2025, the Omni 3 series in April 2026, and the flagship OmniHand 3 Ultra-M at ICRA in June 2026. That launch cadence is reinforced by organization-building signals on the same page, including a 60+ person team scaling toward 100+ by Q4 2026. Public volume claims also move quickly upward, from more than 8,000 OmniHand deliveries by June to more than 15,000 dexterous hands by September 30 and more than 20,000 products shipped on the homepage snapshot. Those numbers are directionally positive for commercialization, but they are not directly comparable because the disclosed scopes shift from OmniHand deliveries to dexterous hands to total products, without separating internal AGIBOT programs from third-party customer units. [CE004, CE012, CE013]

5.4 Technical diligence should focus on SKU reconciliation, reliability, and trust surfaces

The largest product risk in the current evidence set is not lack of activity; it is the gap between rich demo and launch narratives and thin operating proof. Public materials do not reconcile the OmniHand 3 family into a clean SKU-level architecture map: independent demo coverage describes OmniHand 3 Ultra as a 21-DOF direct-drive hand, while APC coverage describes OmniHand 3 Ultra-T as a 22+3-DOF tendon-driven system. That may reflect different products rather than contradiction, but it still means buyers need exact SKU-level diligence on actuation, sensing, and supported workloads. Just as important, the assigned pool does not establish production reliability metrics, customer-visible support SLAs, or external trust/compliance documentation. For underwriting, those omissions matter more than another hero demo, because they determine whether Agilink is selling a repeatable subsystem or a still engineer-heavy integration project. [CE009, CE011, CE014]

5.5 Exhibits

6.1 Segment coverage is broad, but payer quality is still inferred more than proven

The best-supported customer story in the assigned pool is segment breadth rather than named-account depth. AGIBOT's storefront positions OmniHand and OmniPicker around "interaction & manipulation," "object handling," and "manipulation research," while the AGILINK about page frames the company as a dexterous-manipulation supplier for both research and industry. That combination supports a buyer map spanning robotics research teams, humanoid or general-purpose robot OEMs, and industrial integrators that need end effectors rather than finished robot platforms. It does not, however, identify the payer mix by segment or show how much volume comes from affiliated AGIBOT programs versus outside customers, so the commercial quality of the installed base remains partly inferred.[CU001, CU002, CU003, CU006, CU007]

6.2 Adoption proof is real, but still light on named third-party production deployments

The adoption trajectory is stronger than a pure concept-stage robotics supplier, but the proof surfaces remain uneven. AGILINK says the program launched its first products in August 2025 and reached a 1,000th-hand production milestone in December 2025, while Baidu Baike reports more than 8,000 cumulative OmniHand deliveries and more than 10,000 gripper shipments by June 2026. Those milestones imply meaningful physical volume, yet the same source set still fails to show named third-party customers with outcome evidence, contract duration, repeat ordering, or deployment counts by geography. In practice, the assigned pool supports "commercial traction exists" much better than it supports "durable, diversified external customer base exists."[CU004, CU005, CU006, CU012, CU013]

6.3 Expansion depends on partner channels while durability and concentration stay open

The independent evidence points to opportunity, but also to procurement friction and concentration risk. Yicai ties parent-company AgiBot's CP Robotics relationship to vertical application development in life sciences, retail, and wellness, which suggests Agilink's commercial expansion may still be mediated through the broader AgiBot ecosystem rather than repeatable stand-alone demand. Humanoids Daily's write-up of Morgan Stanley's China humanoid survey reinforces that caution: interest is high, but satisfaction is low and only a small minority are already evaluating or launching pilots. The same diligence challenge appears in discovery itself, where unrelated software and cable companies surface under the "Agilink" name, making external customer verification noisier than the headline shipment story. With no disclosed NRR, GRR, churn, renewal, or top-customer mix in the assigned pool, durability and concentration remain material gaps.[CU008, CU009, CU010, CU011, CU012, CU013]

6.4 Exhibits

7.1 Residual risk ranking

The strongest retained evidence points to three risks that can break the investment case if they remain unresolved: parent/control concentration, deployment reliability, and valuation discipline. Official history says Agilink was formed within Agibot, while Baidu reports an Agibot-linked shareholder still controls about two-thirds of the equity, so governance independence is not yet proven by public materials. Product evidence is also still closer to integration guidance, partner workflows, and showcase demos than to audited field reliability. Finally, Crunchbase's unicorn framing arrived within months of the spinout, but the official public package reviewed here still does not disclose the external-customer revenue quality needed to underwrite that price. Kill criteria should therefore focus on control rights, field failure data, and external-customer economics rather than on launch cadence alone.[CR003, CR006, CR009, CR016]

7.2 Governance and regulatory verification risk

Public governance evidence is directionally useful but incomplete. Agilink's own history says the program started within Agibot and only became an independent company in January 2026, while Baidu reports that an Agibot-linked shareholder still owns approximately 67 percent of the shares even after new outside investors joined. That combination leaves meaningful residual risk around board control, IP assignment, related-party transactions, and transfer-pricing. Separately, the retained public set for this chapter did not surface a named package of licenses, certifications, litigation outcomes, or enforcement clearances. That is not proof of a problem, but it is also not enough to clear legal or regulatory exposure; the right next step is a direct diligence package from counsel and management rather than inference from marketing pages.[CR001, CR002, CR003, CR014, CR015]

7.3 Product integration and commercialization execution risk

The product stack looks technically ambitious, but the retained sources describe meaningful integration complexity. AGIBOT's own store says compatibility depends on the host robot's mount, payload, power, communication, and software interfaces, while MANUS explains that OmniHand control can require retargeting, inverse kinematics, and joint-coupling translation rather than a direct joint-to-joint mapping. Those facts raise the bar for customers to achieve stable deployments and make reliability evidence more important than conference demos. The highest-visibility independent-looking article in the pool is still an Agilink-sponsored IEEE Spectrum piece centered on balloon twisting at ICRA, and IFR's broader market context says real-world humanoid capabilities remain concentrated in demonstrators or pilot projects. The residual question is not whether the demos are impressive, but whether Agilink can convert them into repeatable uptime, service, and renewal outcomes.[CR004, CR005, CR006, CR007, CR008, CR009]

7.4 Financing and dependency risk

Financing momentum is a strength, but it also raises price-discipline and process-maturity risk. Crunchbase says Agilink reached a $1 billion valuation after four rounds since the January 2026 spinout, while the official about page simultaneously points to a 60-plus-person team scaling toward 100-plus by Q4 2026 and a dense launch timeline. That combination can be positive if demand quality is proven, yet it also means quality systems, partner management, and customer support must scale rapidly. The public package reviewed here still does not disclose audited revenue, margin, or external-customer concentration data, so investors cannot yet tell how much of the story depends on Agibot internal demand, ecosystem partners, or forward-looking market positioning. A disciplined underwriting posture should therefore tie valuation comfort to verified external revenue quality and concentration rather than to unicorn status alone.[CR010, CR011, CR012, CR013, CR016]

7.5 Exhibits

8.1 Valuation context and price discipline

The public price anchors are strong enough to establish that Agilink has already crossed the unicorn threshold, but they are not strong enough to prove the current price is attractive. Dealroom shows a June 2026 financing event of $138M at a reported $1.4B valuation, while Crunchbase and company- adjacent sources consistently place the business in the May-to-June 2026 unicorn cohort. That combination means the valuation question is no longer whether the market will pay $1B-plus for the asset; it is whether outside investors today have enough evidence on revenue quality, dilution, and repeatable demand to justify following that mark without a discount or stronger protections.[CV002, CV003, CV006, CV007, CV012, CV015]

8.2 Thesis and anti-thesis

The bullish case starts with evidence that Agilink is more than a concept deck: official product pages show live dexterous-hand and gripper offerings, the company says the spinout launched scaled product lines in 2025 and 2026, and independent reporting ties the current financing to deployment-oriented hardware and data programs. The anti-thesis is that the most valuation- sensitive variables remain hidden. Public sources do not verify revenue, gross margin, repeat-order behavior, or preference structure, and the broader humanoid ecosystem still faces commercialization and safety frictions that can compress multiples faster than narrative-driven private marks can recover them.[CV001, CV004, CV005, CV008, CV009, CV010, CV011, CV013]

8.3 Scenarios, comparables, and exit readiness

Scenario framing should stay milestone-based instead of pretending the public record contains a clean revenue multiple. Dealroom and Crunchbase imply Agilink is being priced alongside newly minted robotics unicorns rather than against disclosed cash-flow fundamentals, so the comparable set is best treated as directional sentiment rather than a precise benchmark. The upside path requires proof that shipment claims translate into repeat customer demand, healthy margins, and broader licensing beyond the Agibot ecosystem. The downside path is straightforward: if policy enthusiasm and category scarcity are doing more work than unit economics, a later round can still be flat, structured, or delayed even if the company keeps shipping.[CV002, CV006, CV008, CV009, CV010, CV011, CV013, CV015]

8.4 Final recommendation and diligence asks

The evidence supports engagement, but not blind price acceptance. Agilink has enough product and financing proof to stay on the live diligence list, and the category backdrop is attractive enough that walking away entirely would be premature. Even so, the chapter does not clear the bar for a buy because current valuation markers are better documented than the economics underneath them. The underwriting discipline should therefore be research- more with a stretched valuation stance: revisit only if management can show audited or lender-grade revenue quality, external-customer mix, cap-table terms, and evidence that commercialization can scale faster than standards, safety, and adoption bottlenecks can slow it.[CV010, CV011, CV013, CV014]