The Map Wars · Part 2
Early in 1984, a Stanford PhD named Barry Karlin got lost driving in the Bay Area. The paper map spread across his steering wheel made things worse. That evening he thought: wouldn't it be nice if someone sitting next to me in the car knew the way?
The thought became a company, and the company became an industry. It was called Navteq — the ancestor of today's HERE, and the most consequential name in forty years of digital mapping. Its story contains the original edition of every commercial law this industry has ever learned, including the most expensive one.
What an Illegal Left Turn Taught an Industry
Karlin partnered with Berkeley engineer Galen Collins. Collins built a prototype quickly — and in testing, the computer instructed Karlin to turn left at an intersection where left turns were prohibited.
From that bug, two men in their twenties derived the founding principle of an entire industry: streets and names are not enough. The database must encode turn restrictions, one-way streets, speed limits — every piece of "local detail." A map's product is not lines; it is attributes. Navteq's later creed — hundreds of field researchers driving every street and recording some 160 road attributes — and the "ground truth" DNA that HERE still advertises today, both trace back to one illegal left turn.
Then came the money hunt. Karlin spent a full year being rejected by virtually every venture firm west of the Mississippi. Everyone conceded the idea made sense in the long run; nobody dared underwrite it — the database was too expensive, GPS too primitive, and route-planning software carried the whiff of artificial intelligence. In the end, the president of Budget Rent A Car introduced him to Chicago database entrepreneur T. Russell Shields, who wrote a $500,000 seed check. In August 1985, Karlin & Collins, Inc. opened in Sunnyvale.
File away a coincidence: Etak — the ancestor of car-navigation hardware, founded across town — also started on $500,000 of seed money, from Atari founder Nolan Bushnell. The industry's two great bloodlines both began with half a million dollars and the same obsession: there should be someone in the car who knows the way.
The Kiosk, and the Invention of a Business Model
Mapping the Bay Area's six counties and hundred-plus cities took two years and $3–4 million: the Census Bureau's digital road network as the base, self-commissioned high-resolution aerial photography for registration, and hired drivers covering every road with Dictaphones, narrating everything they saw. GPS wasn't ready, so they built dead reckoning from a gyrocompass and a distance sensor.
The map was finished — and there was no market. Digital maps were too new; nobody knew what to buy them for. So the company built its own endpoint: DriverGuide, a $12,000 countertop kiosk placed in car-rental offices and hotel lobbies, printing driving directions at fifty cents a sheet, the company keeping a quarter. Eighty machines in a year.
By the early nineties, Karlin had understood two things: nobody would truly buy car navigation until the whole country was mapped; and every dollar spent on machines was better spent on the database. So the company pivoted — stop selling hardware, license the database, let others build the products.
Mark the moment: the business model the entire digital-map industry still runs on — data licensing — was invented inside the failure of a direction-printing kiosk. The industry's most valuable lesson cost eighty machines nobody remembers.
Losing $17 for Every $1 Earned
What follows is the original file on this industry's economics.
Philips began investing in 1989, intending to source data for its own car-navigation hardware. Karlin, unwilling to wait out a payoff with no visible date, sold his stake to Philips and left; Collins and most early employees left too. The man who stayed was Shields, the original seed investor, who became CEO and moved headquarters to Chicago in 1996.
The ledger is startling: 1994 — revenue $1.7 million, loss $41.1 million. 1995 — revenue $3.3 million, loss $56.9 million. Seventeen dollars lost for every dollar earned. Philips ultimately sank roughly $600 million into the company and abandoned its own navigation hardware plan, keeping only the ever-growing map. A 1996 IPO attempt was thrown back by the market.
In 2000, Philips recruited an unexpected CEO: Judson Green, a Disney veteran — CFO, then chief of the theme-park business, where he launched the cruise line, Animal Kingdom, and the Hong Kong park approval. Putting a theme-park man in charge of a map company had its own logic: both businesses pour a fortune into one-time infrastructure, then charge for every visit.
Green arrived just in time for the dot-com crash; the IPO was shelved again. But demand finally showed up: in-dash navigation scaled, internet mapping exploded, and revenue climbed from $51 million in 1999 to $272.6 million in 2003. In 2002 the company turned its first profit — seventeen years after founding.
That is the price of admission to the map industry: six hundred million dollars and seventeen years. It is also why, for forty years, this industry has never had more than two or three global players.
The Only Patron Who Ever Made Money
In February 2004 the company renamed itself NAVTEQ; in August it went public at $22 a share, raising $880 million. Philips cut its stake from 83.5% to 41.2%, collecting $783 million, and sold the remainder into the rising market over the following two years.
Fifteen years of patience finally had its answer. And note the identity: of all the strategic patrons who ever wrote big checks in this industry — Philips, Nokia, TomTom, the German OEM consortium — exactly one exited whole and in profit: the one who sold before the top of the cycle. Frame that rule.
A Perfect Bull Case, and Its Ending
At IPO, Navteq held roughly 75% of the market; its only serious rival was Tele Atlas. In 2003, just 3% of new cars in the United States carried navigation, under 10% in Europe. Monopoly share, penetration about to inflect, entry barriers priced at $600 million and seventeen years — no textbook could write a better bull case.
The logic all came true. Penetration took off; the stock climbed from $22 until, on October 1, 2007, Nokia bid $78 a share — $8.1 billion, the highest price ever paid in the history of the map industry, unbroken to this day and likely forever.
Then the logic kept coming true, and the money vanished anyway. In January 2010, Nokia made navigation free — using Navteq's asset to carpet-bomb the paid-navigation market that Navteq's own customers, the Garmins of the world, lived on: the supplier's new owner nuking the supplier's customer base, the most twisted scene in the industry's history. Windows Phone collapsed, write-downs cascaded, the Navteq brand was retired in 2011, renamed HERE in 2012, and sold to a German OEM consortium in 2015 for €2.8 billion. From $8.1 billion to €2.8 billion in eight years.
Being right about the market and buying the asset well are two different things. Every line of the 2004 bull case came true, and the buyer at the top still lost seventy percent. That lesson is worth more than any valuation model.
A Parabola
Line up forty years of price tags: a $500,000 seed (1985) — Philips' cumulative $600 million (1989–2000) — a ~$2 billion IPO valuation (2004) — Nokia's $8.1 billion (2007) — €2.8 billion (2015) — an asset awaiting disposal (2026).
A perfect parabola. The apex's timing was not chance: October 2007 was the first autumn after the iPhone — the eve of the paper map's problem being solved for good, by someone who would never again need to pay for a map. The asset was bought at its all-time high at the precise moment the problem it solved was about to be redefined.
With that, the family tree of this industry can be closed: Etak (a sailor's half million), Navteq (a lost doctor's half million), Tele Atlas (a PC in Ghent) — three origins, three dead names, three parabolas; and the assets all still alive: Etak's blood in Tele Atlas, Tele Atlas's blood in today's Orbis, Navteq's blood in HERE, waiting for its next host.
Navteq left the industry two lessons. The first, at its origin: detailed attributes are the product — everyone learned that one. The second, at its apex: the patron who exits before the parabola turns is the only winner — that one, HERE's current shareholders are still retaking.
Sources: International Directory of Company Histories (Vol. 69, St. James Press) NAVTEQ corporate history; Navteq corporate archives and IPO filings; Nokia's 2007 acquisition announcements; public records of HERE's corporate lineage.
《地图战争》系列 · 第2篇
1984年初，一个叫Barry Karlin的斯坦福博士在湾区开车迷路了。方向盘上摊开的纸质地图帮了倒忙。当晚他想：要是车里坐着一个认路的人该多好。
这个念头后来变成一家公司，公司后来变成一个行业。它叫Navteq——今天HERE的前身，四十年数字地图史上最重要的名字。它的故事里藏着这个行业每一条商业规律的原始出处，包括那条最贵的。
一个非法左转教会行业的事
Karlin找到伯克利工程师Galen Collins合伙。Collins很快做出原型，但试用时电脑指挥Karlin在一个禁止左转的路口左转。
两个年轻人由此悟出了整个行业的立身之本：光有街道和路名不够，数据库必须收录转向限制、单行线、限速——一切"本地细节"。地图的产品不是线条，是属性。后来Navteq"雇几百名外业员开遍每条街、记录160项道路属性"的实测信仰，HERE引以为傲的ground truth基因，源头就是这个非法左转的bug。
然后是找钱。Karlin花了一整年，被密西西比河以西几乎所有风投拒绝——所有人都承认这个想法长期看有道理，但没人敢赌：数据库太贵，GPS还太原始，路径规划软件带着人工智能的色彩。最后是Budget租车公司的总裁把他引荐给芝加哥数据库企业家T. Russell Shields，拿到50万美元种子钱。1985年8月，Karlin & Collins公司在Sunnyvale开张。
顺便记一个巧合：同城的Etak——车载导航硬件的始祖——种子钱也是50万美元，来自Atari创始人Bushnell。这个行业的两大血脉，都始于50万美元和同一个执念：车里该有个认路的人。
问路亭，和一个商业模式的发明
测完湾区六个县一百多个城市，花了两年、三四百万美元：人口普查局的数字路网打底，自拍高分辨率航照做配准，雇人拿着录音机把每条路开一遍、口述所见的一切。GPS还没成熟，就先用陀螺仪加测距做航位推算。
测完了，却发现没有市场——数字地图这个东西太新了，没人知道拿它干什么。只好自己做终端：DriverGuide问路亭，一台1.2万美元的柜台机器，摆在租车行和酒店大堂，打印驾驶路线，每份收5毛钱，公司分两成半。一年铺了80台。
到九十年代初，Karlin想明白了两件事：车载导航在全国地图测完之前没人会真买；做机器的每一块钱，都不如拿去建库。于是转型——不卖硬件，改为授权数据库，让别人去做产品。
请记住这个时刻：整个数字地图行业沿用至今的商业模式——数据授权——是在问路亭的失败里发明出来的。行业最值钱的一课，学费是80台没人记得的机器。
每挣1美元，亏17美元
接下来是这个行业经济学的原始档案。
1989年，飞利浦入局投资，本意是给自家车载导航配数据。Karlin不愿意等一个遥遥无期的兑现，把股份卖给飞利浦离场；Collins和多数早期员工也走了。留下来的是当年给种子钱的Shields，他接任CEO，1996年把总部迁到芝加哥。
账本触目惊心：1994年营收170万美元，亏损4110万；1995年营收330万，亏损5690万——每挣1美元，亏17美元。飞利浦前后砸进约6亿美元，自己的导航硬件计划反而放弃了，只剩下这张越测越大的图。1996年想IPO套现，被市场打了回来。
2000年，飞利浦请来一位意想不到的CEO：Judson Green，迪士尼的老将——从CFO做到主题公园业务掌门人，任上搞出了迪士尼邮轮、动物王国和香港迪士尼的批文。让管乐园的人来管地图，逻辑其实通：两门生意都是先砸重金建一次性基础设施，然后靠每一次到访收钱。
Green赶上互联网泡沫破裂，IPO再度搁置。但需求端终于来了：车载导航起量，互联网地图爆发，营收从1999年的5100万美元涨到2003年的2.726亿。2002年，公司第一次盈利——距离创立，17年。
这就是地图行业的入场券价格：六亿美元加十七年。也是为什么四十年来这个行业始终只有两三个全球玩家。
唯一赚到钱的金主
2004年2月，公司改名NAVTEQ；8月，以每股22美元IPO，募资8.8亿美元。飞利浦持股从83.5%降到41.2%，套现7.83亿，剩余股份在随后两年的股价上行中出清。
十五年的坚持，终于有了交代。而且请注意这个身份：地图行业四十年，所有掏过大钱的战略金主里——飞利浦、诺基亚、TomTom、德系车厂财团——全身而退还赚了钱的只有一个，就是那个在周期顶部之前卖掉的。这条规律值得裱起来。
完美的多头故事，和它的结局
IPO时的Navteq手握约75%的市场份额，唯一的严肃对手是Tele Atlas。而2003年美国新车的导航渗透率只有3%，欧洲不到10%。份额垄断加渗透率起飞加进入壁垒高不可攀——教科书都写不出更好的多头逻辑。
逻辑全对了。渗透率确实起飞，股价从22美元一路上行，2007年10月1日，诺基亚以每股78美元、总价81亿美元发出收购——地图行业历史最高价，至今没有被打破，大概率永远不会被打破。
然后逻辑继续对，钱却没了。2010年1月，诺基亚宣布导航免费——用Navteq的资产，核爆了自己客户Garmin们赖以为生的付费导航市场，供应商的新主人亲手炸掉了供应商的客户群，行业史上最拧巴的一幕。随后Windows Phone崩盘、连番减值，2011年Navteq品牌注销，2012年改名HERE，2015年以28亿欧元卖给德系车厂财团。81亿美元到28亿欧元，八年。
看对市场和买对资产，是两回事。2004年那份多头逻辑，每一条都兑现了，顶部的买家还是亏掉了七成。这一课比任何估值模型都值钱。
一条抛物线
把四十年的价签排成一行：50万美元种子（1985）——飞利浦累计6亿（1989–2000）——IPO约20亿美元市值（2004）——诺基亚81亿美元（2007）——28亿欧元（2015）——待处置资产（2026）。
一条完美的抛物线。顶点的时间不是偶然：2007年10月，正是iPhone上市后的第一个秋天——纸质地图的问题被彻底解决的前夜，而彻底解决它的人，不再需要付费地图。资产被以历史最高价买走的那一刻，恰好是它所解决的问题即将被重新定义的那一刻。
至此，这个行业的家谱可以合上了：Etak（帆船手的50万）、Navteq（迷路博士的50万）、Tele Atlas（根特的一台PC）——三个源头，三次名字消亡，三条抛物线；而资产至今都活着：Etak的血在Tele Atlas里，Tele Atlas的血在今天TomTom的Orbis里，Navteq的血在HERE里，等着下一个宿主。
Navteq给行业留下了两课。第一课在起点：细节属性即产品——这一课所有人都学会了。第二课在顶点：在抛物线拐弯之前退出的金主，才是赢家——这一课，HERE今天的股东们还在补考。
数据来源：International Directory of Company Histories（Vol.69, St. James Press）收录的NAVTEQ公司史、Navteq公司档案与IPO文件、诺基亚2007年收购公告、HERE公司沿革公开资料。