Every company is running three businesses at once — whether it knows it or not. Horizon 1 is the core: the current business that generates today's revenue and profit. Horizon 2 is the emerging: new ventures gaining traction but not yet at scale. Horizon 3 is the speculative: seeds, bets, experiments that won't pay off for years — if they pay off at all. Mehrdad Baghai, Stephen Coley, and David White laid out this framework in The Alchemy of Growth (1999) after studying growth patterns across dozens of companies at McKinsey. Their finding: the companies that sustained above-average growth for a decade or more managed all three horizons simultaneously. The ones that didn't — even the ones posting strong current results — eventually stalled.
The intuition is simple. H1 generates the cash. H2 builds the future revenue engine. H3 creates the options that become H2 in five years. Each horizon operates on a different time scale, requires different metrics, and demands different management. H1 is about execution — margins, efficiency, market share. H2 is about scaling — customer acquisition, unit economics, product-market fit. H3 is about learning — hypothesis testing, technical feasibility, strategic positioning. A team that runs H3 like H1 kills every experiment with premature ROI demands. A team that runs H1 like H3 bleeds cash while "exploring."
The trap is universal and predictable. Companies over-invest in H1 because it generates today's revenue, and every quarterly earnings call rewards it. They starve H2 because it's "not ready yet" — still losing money, still finding its footing, still a distraction from the core. They ignore H3 entirely because it looks like waste. The CFO can't model it. The board can't track it. The CEO can't explain it to analysts. So H3 gets killed in the budget cycle, H2 gets underfunded, and H1 gets all the resources — right up to the moment when H1's market shifts and the company has nothing behind it.
Kodak is the canonical failure. Its H1 — film photography — generated $16 billion in peak revenue in 1996. Its H2 — digital cameras, which Kodak invented — was starved of resources because every dollar invested in digital cannibalised film margins. Its H3 — imaging software, online photo sharing — was explored in labs but never given a path to scale. When digital disruption arrived, Kodak had a monopoly in a dying business and no viable successor. It filed for bankruptcy in 2012. The company didn't lack technology or talent. It lacked the willingness to fund all three horizons simultaneously when the returns on H2 and H3 threatened H1's economics.
Section 2
How to See It
The Horizons Framework reveals itself in how organisations allocate resources across time horizons. The signal is not strategic intent — every CEO claims to be investing in the future. The signal is actual resource allocation: where do the dollars, the headcount, and the A-players go?
Technology
You're seeing the Horizons Framework when Amazon runs AWS as a $100+ billion H1 generating 60% operating margins, invests billions in Amazon Pharmacy and One Medical as H2 bets on healthcare disruption, and simultaneously funds Project Kuiper — a 3,236-satellite low-Earth-orbit broadband constellation — as an H3 option on global internet infrastructure. Bezos structured Amazon to tolerate the P&L drag of H2 and H3 because he understood that AWS itself was once an H3 experiment that became the most profitable cloud business on Earth.
Enterprise Software
You're seeing the Horizons Framework when Microsoft under Satya Nadella runs Office 365 and Windows licensing as H1 cash machines, scales Azure into a $60+ billion H2 business competing for cloud dominance, and seeds H3 bets across AI (the $13 billion OpenAI partnership), mixed reality (HoloLens), and quantum computing (Azure Quantum). Nadella's restructuring was horizon rebalancing: Steve Ballmer had over-invested in H1 (Windows everywhere) and neglected H2 and H3 until Microsoft was irrelevant in mobile, social, and cloud. Nadella didn't invent the technologies. He reallocated capital.
Consumer & Social
You're seeing the Horizons Framework when a company with a dominant product refuses to rest on it. Meta runs Facebook and Instagram ads as H1 — $135 billion in 2023 revenue. It scales Reels and AI-driven content recommendations as H2 plays defending against TikTok. And it invested over $46 billion in Reality Labs between 2019 and 2023 as an H3 bet on spatial computing. Wall Street punished the H3 spending — Meta's stock dropped 64% in 2022 partly over Reality Labs losses. Zuckerberg kept funding it. Whether Reality Labs succeeds is uncertain. That Zuckerberg funds it anyway is the framework in action.
Semiconductors
You're seeing the Horizons Framework when NVIDIA runs gaming GPUs as the original H1 that funded a decade of data centre investment, scales AI accelerators into a new H1 generating $47.5 billion in data centre revenue by fiscal 2024, and seeds H3 bets on autonomous driving (DRIVE platform), robotics (Isaac), and digital twins (Omniverse). Jensen Huang funded CUDA and data centre chips for years while they generated minimal revenue. The gaming division's cash flow covered the losses. By 2024, the H2 bet had become the most valuable H1 in semiconductors.
Section 3
How to Use It
Decision filter
"For every resource allocation decision, ask: does this strengthen H1's current performance, scale H2's emerging business, or seed H3's future options? If 100% of your resources flow to one horizon, you are building on a single time frame — and single time frames are fragile."
The ratio is not fixed. A startup is 90% H1 (survive, hit product-market fit) and 10% H3 (small bets on adjacent possibilities). A mature company generating strong cash flow should run closer to 70/20/10 — 70% on H1, 20% on H2, 10% on H3. The exact split depends on how fast the core market is shifting and how much cash the company generates. The ratio that never works: 100/0/0.
As a founder
The biggest danger is treating your company as a single-horizon business. In the early years, survival demands total focus on H1. But the moment the core business generates positive cash flow, the clock starts. If you don't plant H2 seeds while H1 is strong, you'll be forced to plant them when H1 weakens — and by then you're fighting for survival instead of building from strength. Bezos started AWS in 2003 while Amazon's e-commerce business was still fighting for profitability. It looked reckless. It turned out to be the highest-return investment in technology history. The key: fund H2 and H3 from H1's cash flow, not from its headcount or management attention. Separate the teams. Different leaders, different metrics, different reporting cadence. An H3 team reporting to an H1 VP will be dead within two quarters.
As an investor
Evaluate whether management runs all three horizons or just the one that produces this quarter's numbers. The strongest signal: does the company invest in projects with negative near-term ROI but significant option value? Companies that spend exclusively on H1 produce steady current earnings and fragile long-term prospects. Companies that balance all three horizons show lower current margins — because H2 and H3 are P&L drains — but higher long-term compounding. Amazon's net margins stayed below 6% for two decades because it continually funded H2 and H3 businesses. The market punished the margins. The compounding rewarded the patience. The stock returned over 180,000% from IPO to 2024.
As a decision-maker
Apply the framework to your career. Your H1 is your current skill set — the competencies that earn your income today. Your H2 is the skill you're actively building that will be valuable in 2-3 years. Your H3 is the domain you're exploring with no clear payoff — reading, experimenting, building small projects. A professional who invests only in H1 skills optimises for today's job market and is blindsided when the market shifts. The accountant who spent 2020-2023 learning AI-assisted analysis (H2) while dabbling in prompt engineering and LLM fine-tuning (H3) is positioned for 2025's market. The one who doubled down on Excel mastery is not.
Common misapplication: Treating the horizons as sequential phases — "first we'll master H1, then we'll build H2, then we'll explore H3." The framework's entire point is simultaneity. Running all three at once is uncomfortable because the metrics conflict: H1 demands efficiency, H2 demands growth spending, H3 demands tolerance for failure. Companies that sequence the horizons instead of parallelising them always arrive at H2 too late.
A second misapplication is using H3 as a branding exercise rather than a genuine investment. Announcing an "innovation lab" with five people and no budget is not H3. Google's Waymo is H3 — $5+ billion invested over a decade with patient capital and genuine technical ambition. A corporate innovation theatre with a ping-pong table and no path to scale is a press release, not a horizon.
Section 4
The Mechanism
Section 5
Founders & Leaders in Action
The leaders who mastered the Horizons Framework didn't just tolerate the tension between investing in today's business and funding tomorrow's experiments. They structured their organisations around it — creating the financial architecture, the reporting cadence, and the cultural permission to run all three simultaneously.
Bezos ran the clearest three-horizon portfolio in corporate history. Amazon's e-commerce marketplace was textbook H1 — $575 billion in net sales by 2023, growing at 12% annually, generating the cash flow that funded everything else. AWS was the H2 that became H1: launched in 2006 as an internal infrastructure project that Bezos opened to external developers, it grew into a $91 billion business by 2023 with operating margins above 30%. Amazon's H3 portfolio was simultaneously enormous and underappreciated: Project Kuiper ($10+ billion committed for satellite broadband), Amazon Pharmacy and One Medical (healthcare distribution), Zoox (autonomous vehicles), and Alexa Fund investments across robotics and ambient computing. Bezos protected H3 investments by structuring Amazon as a collection of two-pizza teams with independent P&Ls. An H3 project didn't need to justify itself to the retail division's margin targets. It needed to demonstrate learning velocity. The Fire Phone — a $170 million write-down — looked like a failure. The team that built it developed the hardware, voice recognition, and cloud integration expertise that produced the Echo and Alexa. The H3 failure was the H2 seed.
Nadella inherited a company that had over-invested in H1 for a decade. Under Ballmer, Microsoft treated Windows as the only horizon that mattered — every product, every strategy, every org chart decision filtered through "how does this help Windows?" The result: Microsoft missed mobile (Windows Phone captured less than 3% share before being killed), missed social, and nearly missed cloud. Nadella's transformation was horizon rebalancing at Fortune 5 scale. He stabilised H1 — Office 365 and Windows licensing continued generating tens of billions — but stopped requiring every initiative to serve it. He elevated Azure from a secondary project to the company's strategic centre (H2), investing over $50 billion in cloud infrastructure by 2024. He seeded H3 bets across AI (the OpenAI partnership, Copilot integration across every product), gaming (the $69 billion Activision Blizzard acquisition), and quantum computing. Microsoft's market capitalisation went from $300 billion when Nadella took over to over $3 trillion by 2024 — a 10x increase driven not by H1 performance but by H2 and H3 investments that Ballmer never made. The operational insight was structural: Nadella reorganised engineering so that Azure and AI groups reported independently from Windows, preventing H1 incentives from strangling H2 and H3 resources.
Huang spent two decades building NVIDIA's H1 — GPUs for gaming — into a $10 billion business. Gaming funded everything. Starting in the early 2010s, Huang made H2 and H3 bets that looked eccentric: CUDA programming framework for general-purpose GPU computing, data centre accelerators for machine learning, and autonomous vehicle chips with no meaningful market yet. The gaming division's cash funded these bets for years while they generated minimal revenue. Wall Street ignored them. By 2023, the H2 bet on data centre AI accelerators had become NVIDIA's new H1 — data centre revenue hit $47.5 billion in fiscal 2024, dwarfing the $10.4 billion gaming segment. NVIDIA's market cap crossed $3 trillion. The H3 bets on autonomous driving (DRIVE), robotics (Isaac), and digital twins (Omniverse) are now the company's H2 portfolio, funded by AI's extraordinary cash generation. Huang didn't pivot from gaming to AI. He ran both horizons simultaneously for over a decade, letting H2 mature while H1 generated the cash to sustain it.
Section 6
Visual Explanation
The Three Horizons Framework — Manage the current business, build emerging ventures, and seed future options simultaneously. The horizons overlap; the transitions are continuous, not sequential.
The three curves overlap by design — the transitions are continuous, not sequential. As H1's value contribution plateaus and declines, H2 should be scaling to replace it, and H3 should be maturing into the next H2. The dotted connections show the pipeline: today's H3 seed becomes tomorrow's H2 emerging business becomes the next decade's H1 core. Companies that let gaps form between the curves — waiting until H1 declines before investing in H2 — arrive too late. The framework's power is in maintaining all three curves simultaneously so the pipeline never empties.
Section 7
Connected Models
The Horizons Framework operates at the intersection of strategy, resource allocation, and innovation management. The connected models below describe the forces that make horizon management difficult, the frameworks that support it, and the failure modes that emerge when companies collapse all three horizons into one.
Reinforces
Strategy vs Tactics
Strategy determines which horizons to invest in and at what ratio. Tactics execute within each horizon. The failure to separate the two is why most companies over-invest in H1: tactical optimisation of the current business feels like strategy because it improves measurable outcomes. Genuine strategy is the decision to allocate 20% of resources to an H2 business that loses money today — a decision that tactical thinking would never produce. The Horizons Framework makes the strategic layer explicit: you cannot claim to have a strategy if 100% of your resources serve a single time horizon.
Tension
Innovator's Dilemma
Christensen's dilemma is the mechanism that kills H1 businesses: a disruptive technology enters at the low end of the market, improves until it displaces the incumbent's core product. The Horizons Framework is the structural defence: by funding H2 and H3 investments that may cannibalise H1, the company pre-empts the disruptor. The tension is that the dilemma makes horizon management psychologically excruciating — investing in the technology that threatens your best business requires tolerating the cognitive dissonance of funding your own disruption. Netflix streaming cannibalised Netflix DVDs. The company that refuses to create this tension internally will have it imposed externally — by a competitor with no legacy to protect.
Reinforces
Capital Allocation Options
The Horizons Framework determines where capital goes across time horizons. Capital Allocation Options determines the best use within each horizon. A company that masters both — allocating the right percentage to each horizon and deploying each horizon's capital to its highest-return use — compounds at rates that single-horizon competitors cannot match. Bezos allocated capital to AWS (H2) when it earned nothing, because the expected return exceeded every H1 alternative. The two frameworks are complementary: horizons set the strategic allocation, capital allocation options set the tactical deployment.
Section 8
One Key Quote
"Failure and invention are inseparable twins. To invent you have to experiment, and if you know in advance that it's going to work, it's not an experiment."
— [Jeff Bezos](/people/jeff-bezos), 2015 Amazon Annual Letter to Shareholders
Bezos embedded this line in a letter explaining why Amazon's H3 investments — many of which failed publicly, including the Fire Phone — were not mistakes but structural requirements. The quote reframes the H3 funding question entirely. Most companies ask: "What is the expected return on this H3 investment?" Bezos asked: "What is the cost of not having any H3 investments?" The Fire Phone produced a $170 million write-down. The team that built it developed the hardware, voice recognition, and cloud integration expertise that produced the Echo and Alexa — a platform generating billions in revenue. The H3 failure was the H2 seed.
The deeper logic: H3 investments are not individual bets to be evaluated on standalone returns. They are a portfolio of options where the value comes from the aggregate, not any single experiment. A venture portfolio with a 90% failure rate and a 100x winner on the remaining 10% dramatically outperforms a portfolio of safe H1 investments. Bezos structured Amazon's H3 spending as a venture portfolio inside a public company — tolerating failures that made quarterly earnings volatile because the option value of the portfolio exceeded the cost of the failures.
Section 9
Analyst's Take
Faster Than Normal — Editorial View
The Horizons Framework is the most intuitive strategy concept that companies still systematically fail to execute. Every CEO understands the idea — invest in today, build for tomorrow, seed the future. The failure is not intellectual. It is structural. Quarterly earnings calls, annual budget cycles, and compensation tied to current-year performance create a gravitational pull toward H1 that requires deliberate organisational architecture to resist.
The companies that execute all three horizons share one structural feature: they separate the funding and governance of each horizon. Amazon's two-pizza teams. Google's X lab. Microsoft's independent Azure engineering org under Nadella. The separation is not bureaucratic preference — it is a survival mechanism. When H2 and H3 report to H1 leadership, they get evaluated by H1 metrics. An H3 project judged on quarterly revenue contribution is dead on arrival. The organisational separation creates protected space for different time horizons to operate on their own terms.
The most underappreciated insight: H2 is where most companies fail. The jump from H3 (small experiment) to H2 (scaled emerging business) requires different investment and different leadership. H3 needs explorers — people comfortable with ambiguity, small teams, undefined metrics. H2 needs builders — people who take a validated concept and turn it into a repeatable, scalable business. Most companies kill H3 projects before they're ready for H2, or promote them to H2 without the resources and leadership the transition demands. Google has killed more promising H2 businesses than any company in tech — Reader, Plus, Stadia, dozens of messaging products — not because the H3 experiments failed but because the H2 scaling execution failed.
The AI era is compressing the horizons. What used to take 5-7 years to transition from H3 to H2 now takes 2-3 years. ChatGPT went from research project (H3) to $2 billion+ revenue business (H2) in under eighteen months. This compression means companies must seed H3 investments more frequently and scale H2 faster than traditional planning cycles allow. The companies running annual strategic planning processes are operating on a cadence designed for a world where horizons moved slowly. That world is gone.
My evaluation heuristic: show me the P&L line items that lose money on purpose. The companies worth owning long-term are the ones with deliberate, funded, money-losing line items — because those are H2 and H3 investments the company has chosen to make despite short-term P&L drag. Amazon reported them for twenty years. NVIDIA funded CUDA and data centre GPUs for a decade before they generated meaningful revenue. Meta reports Reality Labs losses explicitly. The absence of any money-losing line item in a profitable company is not discipline. It is a sign that every dollar is being optimised for the present — and the future is unfunded.
Section 10
Test Yourself
The scenarios below test whether you can diagnose horizon imbalance — the gap between what a company claims about its future investments and what its actual resource allocation reveals. The most common analytical error is taking the CEO's narrative at face value. Every annual report claims multi-horizon investment. The budget tells you whether it's real.
Follow the money and the A-players. Strategic decks mean nothing. Budgets and org charts reveal everything.
Is the Horizons Framework being applied correctly?
Scenario 1
A $5B revenue software company allocates 92% of R&D spending to its core product (incremental features, bug fixes, performance improvements) and 8% to a twelve-person 'skunkworks' team exploring three early-stage AI projects. The CEO tells analysts: 'We are investing aggressively in AI and the future of our platform.'
Scenario 2
A consumer electronics company with a dominant smartphone business (H1) launched a wearables division three years ago (H2). Wearables has grown 40% year-over-year to $2B in annual revenue but burns $300M annually. The board debates cutting wearables losses to improve consolidated margins. Smartphone growth has slowed to 3%.
Scenario 3
A pharmaceutical company allocates 60% of R&D to improving existing drugs (line extensions, new formulations), 30% to Phase II and III pipeline drugs (3-5 years from market), and 10% to early-stage university research partnerships. Two of its top three drugs face patent expiry within four years.
Section 11
Top Resources
The Horizons Framework literature starts with the original McKinsey work and branches into innovation management, corporate strategy, and the challenge of managing ambidextrous organisations. The practitioner letters from Bezos and Nadella provide the clearest evidence of the framework executed at scale.
The original source. Baghai, Coley, and White document the Three Horizons Framework based on McKinsey's study of sustained-growth companies. The core thesis — that companies must manage a pipeline of businesses at different maturity stages simultaneously — remains the definitive articulation of why single-horizon management leads to stall-outs.
The most detailed public record of three-horizon management at scale. Bezos documents the logic behind AWS (H2 to H1 transition), the Fire Phone failure (H3), Prime Video (H2), and Project Kuiper (H3) — explaining why each investment was made and what metrics governed it. The 1997 letter establishing "Day 1" thinking is the philosophical foundation. The 2015 letter on failure and invention is the H3 manifesto.
Moore maps innovation types to the Three Horizons: incremental innovation serves H1, adjacent innovation serves H2, and transformational innovation serves H3. His contribution is the practical guidance on managing each type — different governance, different metrics, different talent — within a single company. Essential for operationalising the framework beyond the conceptual level.
O'Reilly and Tushman coined "organisational ambidexterity" — the ability to exploit current businesses while exploring new ones. Their research demonstrates that structurally separated units with distinct cultures, processes, and metrics outperform companies that integrate exploration and exploitation within a single organisation. The academic evidence for why H2 and H3 teams must operate independently from H1 leadership.
Nadella's account of rebalancing Microsoft's horizons from Windows-centric H1 fixation to a multi-horizon portfolio spanning cloud, AI, and gaming. The book documents the organisational, cultural, and strategic changes required to shift a $300 billion company from single-horizon management to three-horizon execution. Particularly valuable for the practical challenges of horizon rebalancing inside a large organisation with deeply entrenched H1 incentives.
H1 is exploitation — extracting maximum value from known opportunities. H3 is exploration — searching for new opportunities with uncertain payoffs. H2 sits at the inflection point where exploration begins converting to exploitation. The multi-armed bandit literature shows that optimal strategies allocate some fraction of pulls to exploration even when exploitation is currently profitable — because the information value of exploration prevents the agent from getting stuck in a local optimum. The Horizons Framework translates this mathematical result into corporate strategy: even when H1 is performing brilliantly, dedicating resources to H3 exploration prevents the company from optimising itself into irrelevance.
Taleb's barbell pairs extreme safety with small, speculative bets — avoiding the middle. Applied to horizons: invest heavily in H1 (predictable cash generation) and take aggressive H3 bets (asymmetric upside), while being selective about H2 (the middle that often produces mediocre returns for substantial investment). Amazon's portfolio resembles a barbell: massive retail and cloud operations (H1) paired with moonshot bets on satellites, autonomous vehicles, and healthcare (H3). The barbell structure protects the downside — H1 cash generation survives even if every H3 bet fails — while preserving the upside that makes long-term compounding possible.
Leads-to
Reversible vs Irreversible Decisions
H3 investments should be structured as reversible decisions whenever possible. An H3 experiment that costs $5 million and can be shut down in six months is a call option — limited downside, uncapped upside. An H3 bet requiring $5 billion in irreversible infrastructure is a different risk entirely. Bezos' distinction between Type 1 (irreversible) and Type 2 (reversible) decisions maps directly to horizon management: H3 should be dominated by Type 2 decisions — fast, cheap, and reversible. H1 can tolerate Type 1 decisions because the business is well-understood. The companies that make irreversible H3 bets or treat H1 decisions as experiments get the risk profile of each horizon wrong.
Scenario 4
A cloud infrastructure company commits $15B over five years to build a quantum computing division from scratch. The CEO hires a Nobel Prize-winning physicist to lead it, gives the division independent P&L authority, and tells the board that quantum revenue will be immaterial for at least seven years. The core cloud business generates $30B in annual revenue with 25% operating margins.
