120: A Return Visit to Insta360's 刘靖康 After Its STAR Market Listing: What Else Could This Be but an Extreme Sport!
Summary
Insta360 listed on the STAR Market on June 11, opened up 285% and briefly surpassed RMB70B in market cap, but 刘靖康 likened the listing to “getting the marriage certificate five years ago and holding the wedding today,” not a change in operating logic. He acknowledged feeling proud “to some extent” of hitting a rare milestone, but also urged the public “not to get harvested,” warning against inflating expectations with concepts or rewriting their view of the company based on short-term volatility; the real new phase is the set of opportunities, challenges and unprecedented possibilities after the listing.
Insta360 is deliberately doubling the competitive stakes: the Ace series is entering the wide-angle action-camera market validated by GoPro and DJI, while DJI is arriving at Insta360’s home turf, where its global share of the 360-camera market is close to 70% and the category contributes more than half of revenue. 刘靖康 believes offense requires finding one gap in an opponent’s barrel, while defense requires filling every gap, meaning an imaging company must become a “12-sided warrior”; the cost may be lower margins, profits and even unit sales, while the payoff is faster improvement in product, marketing, channel and organizational capabilities under pressure from high-level competitors.
刘靖康 sees imaging as more like gaming than autonomous driving: the space for defining both demand and outcomes remains open enough that competition may not end in homogenized specs and a price war. Autonomous-driving standards around safety, speed and comfort converge tightly, while imaging can keep diverging through camera position, movement, filters, editing, effects and subject matter—“no girl complains that her clothes come in too many styles”; that leaves room for multiple companies to make money, but also means consumer electronics usually has a threshold rather than a true moat, so no leader can rest easy.
Insta360 has built an exceptionally thick survival cushion for active competition: its cash on hand could keep roughly 2,000 people operating for five years with “not a single unit of revenue.” 刘靖康 treats the worst non-fatal operating losses as tuition for capability building, based on his view that big opportunities usually go to the player that is most accumulated and ready from the prior cycle; he formed that framework after visiting companies in Taiwan and Japan in 2019 and cites Xiaomi and Huawei entering autos to argue that new businesses must inherit a sufficiently broad base of old capabilities, rather than relying on “throw ten things at the wall and one will work.”
Product expansion is not simply trend-chasing: it moves from replacing existing solutions and increasing penetration through new form factors to validating demand that has never been proven. The Ace line first attacked pain points such as night shooting, while Ace Pro 2 addressed wind noise and pre-recording; the street-photography kit used DIY signals, research and small autonomous decisions to make “old trees bloom again.” Higher-risk hardware-software products may disappoint their first paying users, but could also buy 1-2 years of lead time; editing software is about to undergo what 刘靖康 calls a “qualitative change,” though he offers only “stay tuned.”
The organization is moving beyond “business first, people second” toward “developing the business by developing people,” with a long-term vision of a university-like company that continuously creates and transmits knowledge. Insta360 has more than 3,000 employees across over 600 job types, and a break anywhere in the chain forces others to fill the gap; 刘靖康 is therefore studying “employees first, customers second, shareholders third,” pushing small decisions downward and allowing tolerable mistakes through “context not control,” while admitting that he still often gives direct answers out of efficiency impulses—“I can make a whole speech about it”—and remains far from the required standard on delegation and emotional management.
The largest non-financial variable in the strategy remains the founder’s sustainability: 刘靖康 holds 8-9 meetings a day and believes the current state cannot last five years, leaving him to hope that team capability improves and decision density falls in roughly two years. He attributes starting a company, going overseas and knowingly entering head-on competition with strong rivals to “YOLO”; the long-term goal is not to become “another DJI,” but to build world-class products and validate a new kind of company founded on human development and creation that can still operate after the founder steps aside. For now, his corrections are to reduce micromanagement and short-term reactions to competition, and return more attention to long-term thinking, family documentation and life.
Deep dive
1. The Listing Was a Delayed Wedding; the Real New Phase Is Just Beginning
程曼祺 opened with Insta360’s June 11 STAR Market listing: the stock rose 285% at the open and briefly topped RMB70B in market cap, making 刘靖康, in her words, the “youngest chairman on the STAR Market.” His first reaction was instead: “We got the marriage certificate five years ago; today we held the wedding,” because the listing review had already cleared in September 2021.
The listing’s significance lies more in YOLO. Listing an A-share company of this type at his age is objectively unusual, and it also means the company may encounter “possibilities that have never occurred on this stage.” He said he naturally felt “proud to some extent,” but not because of the “youngest” label.
刘靖康 sees life as a process that can only be changed once: “You arrive with nothing, and you leave with nothing.” Completing old goals, doing what others have not done and entering an unknown new phase are what excite him.
His first public message to the secondary market was unusually cautionary: “I urge everyone not to get harvested.” Investors should not let concepts inflate their expectations for the company or overreact to every swing in the stock.
2. In Two Years, the Business Expanded Horizontally and Deepened Vertically—and the Organization Now Covers the Full Lifecycle
刘靖康 summarized the business changes in two directions: “As the business becomes broader, it is also becoming more vertical.” Horizontally, the company is expanding product lines, single-line SKUs and price bands; vertically, it is moving from easy-to-measure online marketing into offline channels and specific verticals in each country, requiring more precise, higher-quality growth.
The deep end no longer permits broad-brush spending. The same marketing budget must generate more units, and the same unit volume must consume less spending; store training, merchandising and reaching segmented user groups have become basic capabilities.
Organizationally, the matrix structure has expanded from product and R&D into commercialization. Virtual teams previously covered project approval, development and launch delivery; they must now also handle product commercialization and full-lifecycle management, while business teams are no longer merely large functional departments.
The third change is adding “people first, business second” to “business first, people second.” For products, marketing and technology without existing answers, the company first builds teams capable of creating original knowledge, then lets those teams determine what to do—“developing the business by developing people.”
3. The “Hunter Searching for Prey” Framework Remains, but Unspoken Demand Is Far Larger Than Expected
The hunter framework from two years ago remains intact: find an existing vertical market, confirm a real pain point and decent gross margins among leading companies, and preferably target a market with sales potential on the order of RMB10B—not so small that it lacks value, but not so large that every giant rushes in.
Such businesses offer two layers of return. Success creates solid revenue and cash flow; moving from last place to the front of the industry is like “leveling up by fighting monsters,” creating transferable capabilities.
The first change is in demand research. Insta360 now uses a more organized and systematic approach to finding pain points. More importantly, the team discovered that “demand does not live only in what customers can articulate,” and that this hidden space is much larger than it imagined two years ago.
刘靖康 compares imaging with games. Users often cannot describe the game they need, but when they encounter a new world, set of rules and gameplay, they recognize it as “exactly what I wanted.” Imaging can likewise use accessories, shooting capability and finished-image style to make an old product bloom again; the Ace Pro 2 street-photography kit selling out is one example.
4. Insta360 Divides Innovation into Replacement, Expansion and Demand Creation
The first layer is replacement innovation: both the scenario and solution are already defined, and a more efficient solution replaces the old one. The Ace Pro series and meeting products fall into this category; selling one more unit may mean a competitor sells one fewer.
The second layer uses form-factor innovation to increase penetration in existing scenarios. A thumb camera is lighter and can enter cycling and running scenarios where GoPro penetration was limited; a 360 camera can raise action-camera penetration among skiers, so incremental sales do not necessarily equal lost GoPro sales.
The third layer is more fundamental: no one has previously proven that users have the need or are willing to pay for it. 刘靖康 cites the Walkman, cameras and game consoles as historical examples, arguing that a genuinely new market comes not from improving a validated solution, but from turning an unproven need into a paid category.
Over the past two years, Insta360 has launched and advanced several projects under the logic of creating new markets, but he did not disclose their specific forms. The public takeaway is simply that the risk is higher, while successful validation would bring greater returns and a longer lead.
5. Ace Is Entering Direct Competition, Not Filling a Gap After 360 Growth Peaked
程曼祺 noted that Ace resembles GoPro Hero and DJI Action in form, use case and user base, and therefore is no longer creating incremental demand from the flank. 刘靖康 confirmed that this is more direct competition, but said different form factors are like sports cars and off-road vehicles: “A sports car cannot do what an off-road vehicle does, and an off-road vehicle cannot do what a sports car does.”
360 and thumb cameras have clear strengths and inevitable limitations. Traditional wide-angle action cameras remain stronger in image quality, battery life and durability, and many scenarios require all 3, making them impossible to replace equivalently with other form factors.
Commercially, even mature formats retain many unresolved pain points. Insta360 identified those problems and believed it could build a better solution, so Ace was not intended to fill a revenue hole; it was an opportunity to gain both commercial and capability returns.
The Ace project was approved in 2021, before its launch in late 2023 and before the high-growth period when X3 sold 1M units in 2023. The original strategy presentation’s “3 circles” described an encircling, differentiated offensive into the existing action-camera market and new scenarios.
6. Night Shooting, Wind Noise and 1-Minute Pre-Recording Show That Mature Action Cameras Still Have Hard Pain Points
When the first Ace Pro entered the market, night shooting was broadly unusable on action cameras. Insta360 set the target of producing good footage in both day and night; some rainy-night motorcycle footage from Tokyo was mistaken in the comments for “a scene rendered by some game.”
Ace Pro 2 continued to address audio capture while riding. At roughly 80 km/h on a motorcycle and 40 km/h on a bicycle, wind noise can overwhelm the engine, gravel striking the frame, chain noise and even rapid breathing that riders actually want to keep. The built-in windscreen foam and AI algorithms isolate voices while preserving specific frequency bands.
Fishing users need pre-recording because they cannot press the shutter before the fish bites. GoPro and DJI at the time covered only about 30 seconds; Insta360 extended the feature to 1 minute, enough to capture most of the full sequence from “lifting the fish to putting the rod away.”
刘靖康’s conclusion is that image quality, battery life and ruggedness are already traditional action-camera strengths, but maturity does not mean every problem is gone. As long as a specific scenario retains a hard need, details such as night shooting, audio and pre-recording can create an entry point.
7. Competition Will Hurt More, but Could Expand Both the Market and the Company’s Basic Capabilities
刘靖康 does not romanticize competition: “The process will definitely be more painful,” while a better outcome is only probable, not guaranteed.
For Insta360, competition will force up marketing efficiency, product cadence, store training, merchandising and R&D specifications. Specs once deemed sufficient may need to be upgraded, while better experiences and customer-acquisition efficiency could in turn expand the total market.
He uses new-energy vehicles as an example. Automakers were more comfortable early on; competition made the process more painful, but every company still sold more units than it did at the beginning. 程曼祺 immediately noted that auto competition is now brutally intense and the supply chain is “suffering badly”; 刘靖康 accepted the counterexample and added the condition for a better outcome.
The condition is that the market cannot converge too far. If all products ultimately reduce to a few common standards, competition will quickly become a price war once participants reach roughly the same level.
8. Imaging Is More Like Gaming Than Autonomous Driving, Where Evaluation Standards Converge
刘靖康 uses autonomous driving as an example of a convergent market: first, can the vehicle arrive safely; next, can it arrive quickly; finally, are acceleration and braking comfortable? The standards are clear and converge layer by layer, so once participants approach one another, price competition is often all that remains.
Games compete for the same pool of user time but are difficult to substitute on an equivalent basis. Different companies can create independent worlds, rules and gameplay, giving users different forms of pleasure; multiple companies can therefore make money and reinvest high margins into developing new titles.
Imaging has the same abundance of variables: where the camera sits, how it moves, which filters and editing techniques are used, what effects are layered on, and whether the subject is a flower, a child or a person all change the result. “Its outcomes have an extremely high degree of freedom.”
程曼祺 asked whether imaging could turn from an open world into a closed one. 刘靖康 said no: scenarios are dispersed and outcomes can be freely defined. He closed with a colloquial analogy: “No girl complains that her clothes come in too many styles.”
9. The “12-Sided Warrior” Cannot Hide a Weakness Behind One Strength Forever
Imaging opportunities come from complex demand and a long supply chain; the risks come from the same source. 刘靖康 says the key mindset is not to underestimate the upside from improving any link or the downside that any link can create.
The market often talks about a “6-sided warrior.” His version is: “To do imaging well, you need to be a 12-sided warrior.” No weakness can be allowed to persist indefinitely; every weakness should instead be treated as a starting point for reaching the industry ceiling.
Early Insta360 360 cameras were often criticized for poor color, while GoPro had a reputation for “Color Science.” 刘靖康 now confidently says that among mainstream and novice users, Insta360’s color has become the best-looking option.
This is also his defensive logic for direct competition. Imaging is unlikely to permanently conceal every other flaw with one outstanding strength; each weakness must be filled, and yesterday’s weakness may become a new competitive lever once fixed.
10. Insta360 Took DJI’s 360 Entry Seriously Early, but Still Moved Too Slowly
DJI’s move into 360 cameras did not come as a surprise. 刘靖康 says the team had reached that conclusion years ago and had obtained a rough timing range from the supply chain; the problem was not awareness, but that “execution was still a little slow.”
The delays included features that might have arrived in X4 had the company acted earlier. Auto-editing could have iterated faster, and the app rebuild could have started sooner. Insta360 paid a heavy price to redo the 360 editing experience at the time; earlier action would have made that quarter far less painful.
X5 launched roughly 1 year after X4. The host noted that previous generations had been about 18 months apart; 刘靖康 said the 1-year gap between X5 and X4 was not because of DJI, while acknowledging that competition would inevitably accelerate iteration, as seen in the changed cadence between DJI Action 5 and Action 4.
This is what 刘靖康 sees as Insta360’s first genuine encounter with a high-level opponent from a defensive position. In phone gimbals and action cameras, Insta360 had mainly been the attacker; now DJI has entered the core battlefield of the X series, reversing the roles.
11. The 360 Competition Has a Multi-Level Outcome Tree; 刘靖康 Refuses to Give a Static Win Probability
The ideal case is an expanding total market and rising Insta360 unit sales. Its previously near-monopoly share would inevitably fall, but gross and operating margins would remain intact.
The next-best case is rising unit sales with lower margins, but no decline in total profit; operating profit and net income could even continue to grow. Further down the tree, total profit could fall, unit growth could remain limited, or both sales and profits could suffer.
刘靖康 refused to select one outcome as his forecast: “There are a lot of dynamics here.” The result depends on what products the competitor launches, what tactics it uses and how Insta360 responds.
Whatever level the manufacturers end up at, consumers benefit. 程曼祺 concluded that he fully understands the brutality of competition; his response was not to deny the risk, but to lay out the variables and loss tiers one by one.
12. The Fogging at Ace’s Launch Was Not “Not Knowing What You Didn’t Know”; the SOP Never Entered the Organization
刘靖康 is “very satisfied and very confident” about Ace’s current stage, but acknowledges an extremely dark period: the launch suffered fogging, thousands of units were recalled in batches, and sales were suspended for months.
He rejects the idea that this was a case of “not knowing what you didn’t know.” The postmortem pointed to inadequate implementation and execution safeguards, especially incomplete execution of effective measures and SOPs. The company had SOPs and new hires knew they existed, but lacked a mechanism to quantify and confirm that every step had been fully executed.
The deeper problem was that new employees knew the process but not necessarily why it existed. Once older employees make mistakes, correct them and codify SOPs, newcomers may still approach the same work as if it were their first time because the knowledge was never fully transmitted.
刘靖康 compares this with an aviation checklist: “Behind every checklist item may be the cost of one life.” A captain must execute the checklist; understanding why each rule exists enables one to generalize and reach a higher level.
13. 10-bit Is 64x 8-bit, but That Does Not Mean Most People Get Better Straight-Out-of-Camera Images
The classic tension in spec competition is 10-bit versus 8-bit. It is a fact that 10-bit has 64x the color space of 8-bit, but 刘靖康 stresses that this does not automatically mean better color; huge numbers of 8-bit videos on Douyin look excellent.
10-bit is valuable for users who color-grade and are willing to post-process, and it raises the ceiling. But most people do not color-grade and do not want to manually process footage that could have looked good straight out of the camera. Offline stores, however, can easily translate “64x” directly into “better color.”
His analogy is: “You have a watermelon knife, I have a dagger, but my martial arts are better, so I can still beat you.” Insta360 chose to first improve straight-out-of-camera color and market the real content users create, so consumers can see the result directly.
刘靖康 says the platform switch will bring an upgrade to 10-bit. Even without upgrading for now, he says the company has confidence. The point is not that 10-bit is useless, but that spec language not yet converted into mainstream experience should not replace product judgment.
14. CMI Turns Demand Insight from Personal Inspiration into a Team System; Firsthand Experience Raises the Ceiling
Insta360’s CMI department studies motorcycles by examining not only camera pain points, but also vehicle types, rider profiles, behavior before and after riding, psychological needs and the full usage chain. The company has also built a systematic framework for judging demand value versus feature value.
刘靖康 describes the shift as moving from individual innovation to team and system innovation. Given the same facts, teams use brainstorming and other methods to generate feature hypotheses, rather than treating one product manager’s intuition as the entire basis.
The company reimburses coaching fees and equipment for employees in key roles who take part in relevant sports, encouraging R&D and marketing staff to become target customers. Two years ago, his emphasis was that one did not need to be a rider to serve riders; the current correction is that professional methods set the floor, while firsthand experience captures details research can miss.
The highest ceiling lies at the intersection of product, technology and target customer. By encouraging R&D and marketing employees to participate in the sport themselves, the company increases the chance that they become target customers.
15. The Product Matrix Is Expanding by Focal Length Because the Mission Covers Life, Not Just Sport
刘靖康 explains product boundaries through focal length. Wide angles capture more of the environment but struggle with distant subjects; 50mm, 80mm and wide apertures better isolate people and blur backgrounds. 360 cameras have a high ceiling for landscapes and documentation, but their natural ceiling for portraits may not match that of mirrorless cameras.
Products such as Mix Air, along with rumored Pocket-like forms, can therefore be understood as combinations of different focal lengths and forms of subject expression. The company aims to cover the scenario mix needed to document life, not merely add SKUs within the existing action-camera boundary.
He puts it directly: “Our mission is to help people better record and share life, not to record and share sport.” If the goal were only sport, there would be no need to expand focal lengths; life requires broader product capabilities.
Asked whether mirrorless cameras are entering a new window, he rejects trend-chasing as a decision rule. When Insta360 entered action cameras, GoPro was already in decline; the sequence remains unmet demand first, followed by commercial returns, capability building and incremental competition.
16. New Businesses Improve Cash Flow, but Defending the Core Battlefield Builds Capability Faster
Commercially, new categories are more tempting: they bring new cash flow, reduce the share of revenue from any one source and improve resilience. In actual resource allocation, however, Insta360 still places defending the core battlefield first.
刘靖康 breaks capability growth into 2 variables: the opponent’s level, and whether the company is attacking or defending. Attacking a weak rival may make money but produce limited capability returns; attacking a strong rival requires only finding one opening, while defending against one means checking every dimension.
His barrel analogy is clear: “If you attack, you find where the gaps are in his barrel and stick a knife in. If you defend, you have to inspect every gap in the barrel and fill it.”
Defense also means extending existing strengths, not just repairing weaknesses, to create new opportunities and moats. That is why he believes defending the X series will be harder than attacking phone gimbals and action cameras, but will also improve capabilities faster.
17. X5 Deliberately Eliminated RMB36M of Annual Accessory Revenue, Making Defense About Experience Rather Than the Ledger
X5 launched at only about RMB300 above X4, while the product cycle shortened from roughly 18 months to 12, reflecting an aggressive cadence under home-market competition. 刘靖康 goes further in describing “more for the same price” as product self-cannibalization.
In the X3 era, users could buy a relatively expensive protective lens separately. X4 made a plastic protector standard and also introduced a tougher glass protector, which generated about RMB36M in annual revenue for Insta360.
With X5, the team built the hardening capability directly into the native lens and changed the glass, deliberately eliminating that accessory demand. The extra weight, image-quality impact and additional payment associated with the protector disappeared as well.
If the old solution damaged the lens, waterproof sealing and adhesive processes made factory repair time-consuming; a single lens cost roughly RMB500-600, and a pair could exceed RMB1,000. The new design lets users buy a lens and tools for a little over RMB100 and replace it themselves. “In every dimension, the experience is better and more intuitive for customers.”
18. Consumer Electronics Has Thresholds but Few Moats; Open Innovation Makes It Harder to Rest Easy
刘靖康 distinguishes between a threshold and a moat: “A threshold is something you can catch up with if you spend enough time; a moat is something you still cannot catch up with after spending time.” Even with close to 70% global share in 360 cameras, Insta360 can hardly claim to be safe.
Strong moats more often come from network effects, two-sided markets or flywheels: content platforms, e-commerce, food delivery, ride-hailing, the App Store and developer ecosystem, the friend network of Xiaotiancai watches, and Tesla’s vehicle-data/FSD experience loop.
Consumer electronics is the opposite. The technology can be difficult, but product differences are often visible; a later entrant may replace a leader with a better experience or lower price. Imaging’s open world creates more room for differentiation, but also makes scale, channels, brand and supply chain difficult to turn into an uncrossable moat.
DJI’s long lead in drones does not prove that the door is welded shut. 刘靖康 explains through compounded probabilities: if 9 of 10 companies have capability in a given dimension, the probability of meeting the bar across multiple dimensions simultaneously remains low; but if a later entrant methodically gets each item right, the door can still open.
19. A 5-Year, Zero-Revenue Cash Cushion Makes Direct Competition an Affordable Tuition Bill
Insta360’s cash reserves are currently sufficient for roughly 2,000 people to survive 5 years with “not a single unit of revenue.” 刘靖康 deliberately clarifies that this means no money coming in at all, not merely no profit.
That determines whether competition is worth entering. If the cash balance supported only 3 months of life, the company would obviously not choose to compete; a long cash runway makes temporary commercial setbacks against a high-level rival bearable.
The best case is simultaneous growth in capabilities, revenue and profits. The next is no growth; the next is earning less and treating the shortfall as “tuition” for capability development. As long as the company survives, capabilities accumulated through competition can pay back in future businesses.
Even if losses expand sharply, the cash reserve permits “a few more hands.” The bet is not that risk does not exist, but that the survival floor is clear and the potential capability returns can be reused across businesses.
20. The Core Judgment Formed in 2019: Big Opportunities Go to the Player Most Ready from the Previous Cycle
After visiting decades-old companies in Taiwan and Japan in 2019, 刘靖康 saw how capabilities deepen and broaden across cycles, making a company the most ready player when the next opportunity appears. “The answer simply appeared on the page.”
He cites OpenAI, Xiaomi and Huawei. Companies that seize a major opportunity and truly build it out are often not blank sheets when the opportunity arrives. Xiaomi’s auto business inherits software, marketing, product experience and supply-chain capabilities; Huawei’s auto business likewise inherits prior-cycle accumulation.
New businesses and old businesses therefore need sufficient continuity. If a company jumps from cameras to robot vacuums, power banks or earbuds, old capabilities cover only a small part of the new business, leaving the decisive gaps to be rebuilt from scratch.
On Dreame’s broad expansion, he first praised its focus on capability building, then expressed reservations: “Trying things is not a very economical or efficient approach.” The more economical path is to establish the success factors first, then accumulate capabilities through monetizable bridge businesses—for example, building phone gimbals before drones.
21. Insta360 Rejected Automotive Cameras Not Because It Missed the Market, but Because It Could Not Control End-Point Value
The company had received B2B proposals to use its 360-related capabilities for automotive modules and in-car cameras for autonomous driving. It ultimately declined, first because the business had little to do with the mission of “recording and sharing life”; the initial motivation was simply a commercial opportunity.
The capabilities were not naturally transferable either. Companies that truly excel at automotive modules had accumulated capabilities over the prior cycle, and Insta360 could not reach the same level quickly. As a non-exclusive component supplier in the middle of the chain, it also could not define end-point value or pricing.
刘靖康 therefore concluded that Insta360 would struggle to see whether such a business could sustain good gross margins. A Tier 1 or Tier 2 supplier extending from its core business may make sense, but a consumer company entering the same chain cannot determine its profit pool through its own capabilities alone.
B2B is not permanently off-limits. If the mission changes, or if the company builds a stable “grain silo” to support better consumer businesses, it would enter under a standard of long-term investment and becoming No. 1 or at least No. 3. The decision must come first; the company cannot start small and decide along the way.
22. The Cinema-Camera Project Taught Insta360 That Preparation Cannot Be Replaced by Blind Trial and Error
4-5 years ago, Insta360 saw certain cinema-camera companies generating several hundred million RMB in annual revenue and assumed it might have the capabilities to compete, so it launched the project around the opportunity. Halfway through, it discovered that the technology, demand and Go To Market were all much more complex than expected.
If the founder did not personally dive in, the company needed a product manager who understood both deep technology and cinema demand, but no ready supply of such people existed in the market. Customers also had extensive non-standard needs, entrenched workflows and preferences for legacy brands.
Patents added another constraint. 刘靖康 cites RED’s patent on in-camera 8K recording: the technology itself may not be difficult, but the patent was enough to block other companies from using an effective solution. The project was later cut, and the loss was not only financial; it also demoralized a team that had invested substantial effort.
The experience made him distrust moving before the plan is settled. If information could have been collected beforehand, laziness or blind confidence should not defer the problem; iteration is superior only when an industry has no precedent and experiments are required to generate incremental information.
23. Real Iteration Updates the Model with New Information; It Does Not Excuse Failure to Think Clearly
New software features are suitable for low-cost iteration. Form a hypothesis and solution, observe the user funnel, workflow, activity and retention, then study why some users drop off at a step while others continue using the product, and use that analysis to guide the next round.
Brand enhancement cuts both ways. Greater trust in Insta360 raises conversion and lowers decision friction; if the company merely adds free software features for people who already bought the core product, failure does limited damage because those users did not pay for the feature separately.
An unvalidated new scenario that requires hardware and software together carries much higher disappointment risk because the first customers must pay. Conversely, if the first attempt works, Insta360 could gain a lead of 1-2 years.
刘靖康 says the company has methods to raise the probability of success, but does not promise to eliminate expectation gaps. The only conclusion he is willing to state publicly is: “The greater the risk, the greater the return.” For genuinely new categories, avoiding disappointment among the first users entirely is unrealistic.
24. Expanding the User Base Requires Both More Efficient AI Distribution and a Different PMF
In reality, about half of action-camera buyers use the product for travel and life documentation rather than sport. But as a share of all smartphone users, this group remains tiny, showing that the use case exists while penetration is far from complete.
The first path is to keep the PMF unchanged and improve the efficiency of finding users. At a ski resort, 1 in roughly 50 people may own an action camera; in a mall, even 1 in 500 may not. Interest-based and AI-driven information distribution now makes it more likely that brands can identify dispersed target users.
The second path is to improve the product so the device outperforms a phone in more travel, family and everyday scenarios. The goal is not to push action-camera messaging to everyone, but to expand the situations in which the device is meaningfully better than a phone.
The team can also test new scenarios with old products. If users buy and continue using them, the company can study why they retain; if users leave because wide angles make people hard to shoot, that gap becomes input for the next generation. Expanding the audience itself becomes part of iteration.
25. The Street-Photography Kit Matters Less for Selling Out Than for Proving Small-Team Innovation Can Scale
刘靖康 had almost no involvement from project approval to launch preparation for the street-photography kit: “I only knew about it when they had finished preparing to launch.” When Insta360 had only 2-3 products before 2020, that could not happen; after expansion, it became an organizational outcome he actively wanted.
Autonomy has a resource boundary. Business lines can decide on features, software and accessories costing hundreds of thousands, RMB1-2M or even a few million; investments of RMB10-20M or RMB50-60M cannot bypass the founder.
The product experience comes from many details working together: exterior design, physical-button feel, vibration and sound feedback, straight-out-of-camera color, Leica tuning and finished-image watermarking. Together they create a sense of “taking a photo” that a phone’s virtual button cannot provide.
More important, the team first picked up multiple signals from users’ DIY modifications, conducted serious research, narrowed the hypotheses and only then acted. 刘靖康’s review concluded that this was not luck, but a reproducible method of “spotting signals—researching and validating—implementing at small cost.”
26. “Context Not Control” Requires Keeping the Right Answer in Your Mouth for the Moment
刘靖康 believes adults mainly change behavior through feedback. Correct decisions followed by positive feedback increase confidence in future decisions; mistakes followed by negative feedback lead to adjustment.
A team’s first independent decision cannot be high quality by default. When the founder knows the answer, he can use coaching and prompting to help the team reach it; announcing the answer directly returns the organization to command and control—execution happens, but learning does not.
For a solution that is correct but imperfect, he sometimes lets it proceed so the team receives mostly positive feedback and a small amount of negative feedback. The cost is an avoidable commercial loss; the benefit is that experience is generated through practice rather than merely hearing a principle.
For a clearly suspicious plan that cannot be disproved beforehand and would cost only a few hundred thousand or RMB1-2M, he may also let it proceed. The team may learn a profound lesson, or prove the founder wrong; both outcomes generate feedback.
27. Large, Irreversible Risks Require an Override; the Founder’s Emotions and Efficiency Impulses Are Harder to Manage
If an error could cause losses of RMB10M or RMB100M, the outcome is irreversible, the cost of correction is high and the final window has arrived, 刘靖康 will override the team. Whoever makes the call bears the consequences; authority and responsibility must match.
An override can damage team motivation and still requires follow-up management. The more common problem arises when an issue first appears: the founder’s first reaction is often emotional, and once expressed publicly, the team automatically treats the issue as more severe, raising the decision threshold next time.
Giving the solution directly is usually fastest; one meeting can end it. Developing the team means letting the other side keep thinking and scheduling another meeting. 刘靖康 admits he is “not very good” at restraining himself and that he can “make a whole speech about it,” because this runs against human nature.
Insta360 uses programs such as the “rain plan” to reinforce positive feedback: “Sunlight cannot reach everywhere, but rain can.” Motivation starts with small iterations to avoid setting expectations too high; management is gradually moving toward public praise and private criticism, with stable emotions and clear, public performance consequences.
28. “Employees First” Is Not a Benefits Slogan; It Is the Production System Required by an Open Market
刘靖康 divides common corporate rankings into several types. Shareholders first, customers second and employees third follows traditional shareholder-return logic; customers first, employees second and shareholders third treats commercial returns as the natural result of first creating customer value.
He is studying whether Insta360 can become an “employees first, customers second, shareholders third” company. Huawei can also be interpreted as employee-first, but its starting point was more about survival and employees partly overlapped with shareholders; Insta360 wants to emphasize employee development.
The reason comes from the business. Imaging is an open market, and customers pay a premium only for something new; that premium is reinvested in R&D, brand and employee development, creating the next round of customer and business value. “You make money by developing new knowledge or solving problems no one else has solved.”
“Employees first” therefore means capability, creativity and knowledge growth—not simply housing, rent subsidies, club budgets or internal dating bonuses. Benefits can improve the experience, but the real organizational challenge is enabling people to continuously create and inherit new knowledge.
29. With More Than 3,000 People and 600 Job Types, Insta360 Resembles a University That Must Fund Itself
Insta360 has more than 3,000 employees across over 600 job types, connected in a long chain of collaboration. If any one link creates a pit, others must fill it. The company must both develop knowledge and determine how prior knowledge is transmitted across teams and to new hires.
That brings 刘靖康 to the university analogy. Universities conduct research and pass old knowledge to the next generation of students; if Insta360 wants to become a world-class imaging company, it must likewise do both development and transmission.
The difference is funding. Universities have tuition, social donations and government research grants; companies rely mainly on operating income, while investor capital carries return expectations. Efficiency therefore matters because knowledge development must close a commercial loop and remain sustainable.
This also explains the shift from “find experienced people to deliver tasks” to “build teams capable of exploring the unknown.” When no one in the industry has done something before, no existing knowledge is available; internal members must create it and turn it into products, marketing and the next generation of organizational capability.
30. Whether Employee Complaints Matter Depends First on the Target Employee and the Problem’s Prevalence
On social-media complaints from former and current employees, 刘靖康 uses a university analogy. Universities have admissions standards and graduation thresholds; Tsinghua would not lower its requirements because someone complained that the scores were too high. But if the employees the company wants to develop complain about chaotic management, it must pay attention.
Paying attention does not mean satisfying every piece of feedback. First identify whether the speaker belongs to the employees the company truly wants to develop, then determine whether the issue is representative among those employees, whether it impairs efficiency or growth, and whether it is urgent or can be solved by trading time for space.
He agrees with 张一鸣’s idea of treating the company as a product to iterate: identify target employees as one would target users, and assess the criticality of problems as one would dissect product feedback, rather than blending every complaint into one emotional conclusion.
Compared with two years ago, 刘靖康 believes his definition of organizational problems is more systematic. The most hidden and damaging issue may not be that “a particular person is incapable,” but that company complexity creates “gap responsibilities” with no clear owner.
31. Many Low-Performance Problems Come from Gaps in Ownership; Replacing People Only Reproduces Failure
When a team repeatedly fails to deliver, managers usually assume the people are the problem: replace them, apply pressure, or encourage them a few times before becoming disappointed, creating a negative feedback loop on both sides.
But as the business broadens, the workforce grows and competition intensifies, new implicit responsibilities and higher standards arise naturally. If the company never tells a department what it must own, employees cannot know by intuition; tasks touching both Department A and Department B are especially likely to become nobody’s responsibility.
Management therefore needs to identify regularly which objectively existing responsibilities are not assigned to a specific department or person, and which old responsibilities now require a higher standard. Otherwise leaders blame departments for failing, while departments believe they completed everything explicitly assigned to them.
Cross-team conflict also cannot be solved simply by taking sides. 刘靖康 advocates separating facts, emotions and opinions: establish the objective event, identify how emotion amplified the conflict, then locate the difference in principles before forming a new way to collaborate.
32. An Annual Cross-Department “Mirror” Turns Organizational Self-Assessment into Multi-Party Verification
At the end of last year, Insta360 had departments evaluate one another. Department A assessed where Department B ranked relative to the industry and companies it had previously encountered, what it did well and poorly, and what it should do more of or raise to a higher standard in the next phase.
Evaluation cannot rely only on subjective scores. It must name a company that performs better in a given area and provide a basis for comparison. 刘靖康 says this is not one mirror, but “many mirrors” supplied by multiple departments that collaborate frequently.
The mechanism currently runs once a year. Its value lies not just in exposing problems, but in whether the company changes, implements and produces results afterward; repeating a questionnaire every quarter without leaving time to correct anything would only consume the organization.
The team implementing new cultural values is itself an example of a gap responsibility. Management found that behavior in some key roles did not match the values, while existing HR capabilities could not deliver the new requirements, so the team had to be rebuilt around the new responsibilities.
33. Information Security, Matrix Alignment and Leadership Development Are Pushing HR into Deep Water with Few Benchmarks
Information security became a higher-standard function only after competition intensified. Previously, IT patched vulnerabilities based on its own understanding; now Insta360 works with 4-5 security companies to conduct penetration and attack tests, scan all vulnerabilities and prioritize remediation.
The matrix organization also imposes unique demands on HR. Goals must align vertically and horizontally; HRBPs must check whether product-line PBC items are reflected in functional departments and whether the goals of first-level department heads are further broken down, preventing quarterly results from failing to add up to a coherent whole.
Training cannot stop at onboarding process courses or scattered leadership lectures. The company must first identify managers worth developing over the long term, define key-role requirements, identify the gaps between people and roles, and then create continuous development plans.
Leadership should not be measured by attendance, reflections or written tests, but by the shape of talent reviews, the number of composite-role employees, subordinate growth and whether someone can immediately take over when a leader leaves. 刘靖康 admits Insta360 is not there yet, and that HR is much harder to benchmark consistently than software.
34. The Hardest Management Task Is Not Designing Systems, but Getting 600-700 Key People to Truly Want to Develop
刘靖康 defines management’s core and hardest task as “inspiring or motivating people.” He does not need to directly develop all 3,000 employees, but at least 300-400 managers and operators, plus high-potential and core employees, for a total of roughly 600-700 people.
If those people grow, they can continue developing the other 2,000-plus employees. The difficulty is that learning, practice and taking risks all require additional effort, and different people have different reasons for making that effort.
Money is the largest common denominator but not the only motivator for senior talent. Some want achievement; some care about leaders’ evaluations; people with F-oriented MBTI profiles may care more about emotional expression; others score as power-driven on SHL tests. “Even praise has to land on the right point.”
刘靖康 can directly motivate at most 20-30 people, who can then influence a dozen or so each. If the company’s main line is knowledge creation, pure material incentives cannot attract everyone required; interest, passion and positive feedback during the process all need careful management.
35. Organizational Complexity Is Not the End State; It Depends on Whether Insta360 Finds a Large Enough Grain Silo
刘靖康 does not believe organizations become infinitely complex. Google’s core business is close to a “tax-collection” model, allowing employees 20% free time and relatively high tolerance to “breed” new businesses that preserve or expand its monopoly; the organization and business model are internally consistent.
Consumer electronics is highly competitive and complex, with little scope for monopoly. Insta360’s current challenge is to keep each business competitive on its own, which requires a matrix, knowledge transmission and fine-grained incentives.
He offers a hypothetical: Insta360 generated more than RMB5B in revenue last year; if it captures a world-class opportunity worth about RMB20B, it may not need to maintain so many businesses or its current complex structure. It could focus on executing that opportunity and then search for the next RMB20B opportunity.
Apple is clearly maintaining market leadership, while Tencent is closer to the state of having a large grain silo. Insta360’s current structure may last a long time, but the founder is not pursuing complexity for its own sake.
36. Glasses Could Rebuild the Internet’s Entry Point, but They Are Not Insta360’s Core Opportunity
刘靖康 acknowledges that he has seen the early form of a product he cannot yet disclose, but refuses to elaborate: only “when it is proven” might he talk about it. Even after a formal launch, he would not immediately call it a massive opportunity, because by the time everyone understands an opportunity, the window is often already closed.
His view of glasses is clearer: dedicated camera glasses are not the kind of major opportunity Insta360 should pursue. Smartphone makers can bind glasses tightly to phones and cover the broadest demand, from ordering food delivery onward; an independent imaging company cannot control the phone ecosystem.
What could truly disrupt the internet is an AI assistant moving the point of demand capture upstream. Users would no longer first decide whether to open Meituan or Ele.me; they would simply say “order food,” and the system would compare both apps before distributing the order based on price, preference and even commercial weighting.
To control the largest entry point, phone makers must design for the greatest common denominator, sacrificing PMF in vertical scenarios. Those gaps still leave room for specialist imaging companies to “drink the leftovers.” But 刘靖康 stresses that Insta360’s opportunity is not glasses, nor is it something that becomes worth hundreds of billions overnight or in a very short period.
37. Eight or Nine Meetings a Day Preserve Decision Density, but 刘靖康 Wants That Number Down in Two Years
刘靖康 holds 8-9 meetings a day covering R&D, business, organizational design, incentives and personnel issues; the listing is not the main reason he is busy. He says plainly: “In its current state, this is unsustainable.”
Compared with 3-4 meetings a day in the past, most meetings are now compressed to 30 or 45 minutes. Participants must bring a proposal, turning the founder’s task into a multiple-choice question rather than a fill-in-the-blank problem he must solve live.
If a half-hour meeting runs to an hour, the organizer is fined more than RMB100; even if 刘靖康 is the one who runs over, the organizer is still fined because the meeting was inadequately prepared. The money compensates the person whose next meeting was delayed.
He expects micromanagement to decline as talent density, processes and team decision-making improve. If the current intensity remains, he “cannot last 5 years”; if decision density falls in 2 years, he is confident he can continue for another 2-3 years.
38. The Bravest Choice Was Not Entrepreneurship, but Voluntarily Entering a Fight He Knew Would Be Direct
刘靖康 lists his early brave decisions: an unsuccessful roadshow in his sophomore year, joining someone else’s startup, starting his own company, using his own money to pay salaries and securing IDG angel investment 1 month before graduation. Going overseas came later.
But he considers the bravest decision of his life to date the choice to invest in businesses despite knowing direct competition was coming. If he had stayed in a corner, built a business worth tens of billions of RMB and maintained a global No. 3 position, he believes he could have lived very comfortably.
His observation of competition is that No. 1 watches No. 2, No. 2 watches No. 1, and nobody watches No. 3. As long as he does not actively take someone else’s slice, his existing capabilities could keep him comfortably at No. 3; moving from last place into the top 3 and then to No. 2 is destined to be uncomfortable.
The driver remains YOLO. Courage does not mean failing to see severe consequences; it means choosing to bear risks already anticipated: “You arrive with nothing, and when you leave, you take nothing with you.” Dealing with risk is itself part of the only life one gets.
39. The Endgame Is Not Becoming “Another DJI,” but Proving a Company Can Create After the Founder Steps Aside
刘靖康’s long-term goal is more concrete than it was 2 years ago: build world-class products and a world-class company, and create a new organizational form centered on learning, knowledge and human development. “Although we have not made a world-class product yet,” the company must set a higher target to motivate itself.
If the goal is merely to exceed DJI in revenue and profit, the result is only “another DJI.” Comparing his personal cash holdings with 汪涛’s would be even less meaningful. More worth pursuing is the ability to face a Huawei-level opponent without starting from “survival first,” while continuing to develop people, create new things and survive.
The company must also retain the same capabilities after the founder steps aside, just as a university does not suddenly become stronger or weaker because its president changes. 刘靖康 sees this as harder than producing a single world-class product, but also more capable of sustaining long-term motivation.
The cost is continuous suffering. Dozens of factors support success and dozens support failure; when attention falls on the negative factors, pessimism follows. Even seeing DJI produce something “particularly impressive” during a holiday can instantly destroy his ability to relax.
40. Family Documentation Added the Target-Customer Identity; Long-Term Thinking Must Reclaim Space from Competition and AI
Having children did not change 刘靖康’s operating logic, but it changed how he experiences products. Children are most adorable at 2 or 3, and the relationship may not remain as close when they grow up, so parents should document them with richer tools now. For the first time, he clearly became a target customer for family documentation.
In sports, he relies on technical understanding and customer research but is not an extreme-sports user. In family documentation, technology, product and firsthand need form a closed loop, so he believes his ceiling for defining such products would be higher.
Looking back at his time allocation, his answer is: “Too much micromanagement, not enough delegation; too much focus on things, too little on human nature; too much work, too little family.” Even seeing his daughter often requires fitting time into the gaps.
The first adjustment will be the balance between competition and long-term thinking. Expanding competition naturally fixes attention on short-term moves, but overreaction distorts the company; the long-term goal is still what leads it out of competition. On AI, he only reveals that editing will receive a “major upgrade” and a “qualitative change,” leaving the audience with 4 words: “Stay tuned.”