You need Actual Intelligence to craft a winning startup strategy in the AI market bonanza

It’s eighteen years since Satoshi Nakamoto, the founder of Bitcoin, unveiled the cryptocurrency. The crypto industry has gone from an object of mockery in mainstream finance to being broadly accepted by regulators, banks and asset managers. In October bitcoin’s market value peaked at $2.5trn.

Into November, and price volatility returned; bitcoin dropped from an all-time high of $126k  to <$70k. For a speculative asset which produces no income and relies solely on sentiment for future capital gains, the lack of a bullish forward view to support price growth is not encouraging.

In a parallel universe, the first week in November saw some stunning financial data announced around AI, as Open AI and Microsoft updated their partnership agreement, and Open AI moved to a for profit entity. Microsoft’s 27% holding is valued at $135bn. It meant the developer of Chat GPT was worth $500bn.

Then Nvidia shares rose after strong results eased 'Al bubble' concerns. CEO Jensen Huang said sales for Nvidia's AI Blackwell systems were ‘off the charts’. Revenue for the three months to October jumped 62% to $57bn, driven by demand for its chips used in AI data centres, and Q4 sales forecasts of $65bn topped estimates. Last month Nvidia became the first company to be valued at $5tn. Its market cap is rising like the Hokusai wave.

Nvidia is seen as a bellwether for the AI boom. It’s at the heart of a web of partnership deals in  the AI stack which has set the tone with the division of expertise; chipmakers like Nvidia design GPUs; AI labs like OpenAI and Anthropic devise the cutting-edge models; cloud scalers like Microsoft and Amazon host the labs’ models. All co-operate where they can in a closed loop providing eye-watering financial scaling.

All except Google, who have gone all in on vertical integration. Google Cloud installs tensor-processing units (TPUs) to train models built by its lab, DeepMind. The models power its own products. Thia do-il-yourself approach is less nimble than Microsoft’s and OpenAI’s mix-and-match strategy. Google has been branded an AI laggard, and the shares of parent Alphabet, have traded at two-thirds the price of Microsoft’s relative to earnings on average since the ChatGPT moment, despite its net profit growing faster.

But in the past four months Alphabet has gained $1trn in market value, more than in the previous two and a half years and its vertically integrated approach is winning customers. Google really has been a machine-learning company since formation with its clever search algorithm. Demis Hassabis, the head of DeepMind, is leading the charge.

The reason for the initial flatfooted inertia was time spent understanding how they could step into the AI flow without hurting its core search business. Now Google is catching up as its ‘superchips’ TPU processors work with the rest of its hardware and software, including Gemini and is turning into a favourite with AI firms - they prize TPUs’ higher energy efficiency compared with Nvidia’s GPUs. Anthropic will buy an additional gigawatt of computing power from Google Cloud, worth $9bn a year.

No surprise that with the rise of generative AI, everyone has jostled to be on the winning team. Since it launched ChatGPT in 2022, OpenAI has been the one to beat. But its dominance is under threat - on November 18 Microsoft and Nvidia, two big backers of OpenAI, threw their weight behind Anthropic and committed to a $15bn investment, a big rival to the maker of ChatGPT, financed to date by Amazon and Google.

Earnings reports from Meta, Alphabet and Microsoft reaffirmed the colossal amounts of money these firms are shelling out for everything from data centres to chips, but Sundar Pichai, the head of Google's parent firm Alphabet, said that whilst the growth of AI investment had been an ‘extraordinary moment’, there was some ‘irrationality’ in the current AI boom. On the same day, Google threw down the gauntlet with a new model, Gemini 3.

Some analysts have compared the surge in AI stocks to the bubble and dotcom boom of the late 1990s. This saw the values of early internet companies soar amid a wave of optimism over the then-new technology before the bubble burst in early 2000.

For users, the more competition the better. It means the AI labs will continue to burn cash. and the more competition there is, the harder it will be for the labs to generate revenues needed to justify the spending splurge, not to mention their stratospheric valuations.

Pichai’s cautionary warning about ’irrationality’ in the AI investment boom was real: I think no company is going to be immune to a correction including us he said. It reminded me of US Federal Reserve chairman Alan Greenspan in 1996, warning of ‘irrational exuberance’ in the market well ahead of the dotcom crash. Pichai said the tech industry can ‘overshoot’ in investment cycles like this.

Pichai said: AI is the most profound technology humanity has ever worked on. It has potential for extraordinary benefits Meanwhile. OpenAI’s Sam Altman speculated that there are many parts of AI that I think are kind of bubbly right now. The AI surge is, in cash terms, the biggest market boom the world has seen. Yet it comes with a big risk.

That is, the dependence of US stock market growth on the performance of The Magnificent Seven - Alphabet, Amazon, Apple, Meta, Microsoft, Nvidia, and Tesla - collectively comprising one third of the valuation of America's entire S&P 500.

Pichai points out that every decade or so come these inflection points: the PC, then the internet, followed by mobile and cloud. Now it's clearly the era of AI. Everyone's talking about the bubble. When will it burst? Are we in a hype cycle?  Here's my take: I think we'll see a correction, but it's going to be minor, this isn't dotcom or crypto scenario.

Dotcom was companies with no revenue and no path to revenue raising billions; Crypto is just a hedge fund style betting market.  The AI correction is going to be a pause for breath, a realignment and check on exuberance, not a collapse. The hyperscalers aren’t stopping any time soon and the power infrastructure build isn’t stopping either. AI companies have real revenue, from real customers and real use cases. The infrastructure being built is generating real returns. Will some companies fail? Yes. Will some projects fail? Absolutely. Will valuations come down? Almost certainly. But the underlying economics are real as we race towards AGI and SGI.

The real lesson to take from Pichai is this: AI’s value isn’t in headlines or market valuations, it’s in simple, everyday mass market cost-effective utility. Every AI integration we build is designed to save time, reduce friction, and add valuable insights. But it’s moving rapidly.

When Gen AI first broke into the mainstream, companies like OpenAI were seen as infrastructure providers, engineers-built applications on top of them, with the AI models acting as the foundation layer of a vast new ecosystem. But today, everyone is climbing further up the stack.

OpenAI’s recent release of Sora2, a consumer-facing app for video generation is a good example. What was once just a raw capability (text-to-video) is now packaged into an end-user experience, competing head-on with startups that thought they’d have room to build applications. Similarly, Anthropic launched Claude Teams, not just offering API access to Claude models but delivering a ready-made productivity suite for enterprises.

So, against this backdrop of the tech goliaths shaping the dynamics of the AI marketplace, how does an AI startup win in a fast-moving market where inflection points can kill a previously winning strategy overnight? It’s a combination of the lessons of David v Goliath and recognising winning in AI requires Actual Intelligence.

As Marc Andreessen says beating an incumbent comes down to whether the startup gets distribution before the incumbent gets innovation. He goes further and says startups should position themselves as the AI-native alternative in underserved  categories of enterprise software.

As an AI product startup, what should your strategy be? A startup isn't launching from the same place as Google. For small businesses, every tech product investment must be measured, practical, and immediately valuable. Here are some thoughts:

Demonstrate tangible outcomes from your AI offering - Improve a niche. Make a narrow market space much better. Take an AI-native approach to the smallest possible feature set that will drive differentiation and customer traction in volumes and at pace.

Many companies are falling into a proliferation of proofs of concept and isolated use cases that deliver modest, localised efficiency returns but fail to scale. Tools get deployed. Demos impress. But outcomes never materialise. Be mindful of handholding early-stage AI adoption with customers to deliver timely and tangible RoI.

Rapidly iterate - add features to grow with your customers. Winning hearts, minds and wallets in a niche is a start, but expanding into adjacent workflows builds durability and scalability and prevents customers from churning to a broader solution. Whilst focusing on the one thing is a key startup principle, AI will rapidly evolve and step into adjacent markets spaces that will move together, and opportunities will be even bigger. Ai will enable convergence of separate systems into one continuous system, be part of that shift.

Show Actual Intelligence. Finding, winning and keeping new customers and building a Minimum Viable Business is your immediate aim, but unlike previous technology waves, Gen AI doesn’t create value through basic adoption. Chasing trends or adopting AI for the sake of AI is a luxury no one can afford. ROI comes from reimagining how a company competes with AI at its core.

This needs strategic selling in how you position your AI chops. Treating Gen AI just as a tool isn’t going to work, you need to show how your AI service redesigns workflows, rethinks operating processes and enhances your customer’s customer experience. Most companies are stuck in AI experimentation, not transformation; real impact requires business redesign, not just tech deployment.

Focus on culture as well as tech Many companies are deep in experimentation, but may have gone to early, stalled and need to reboot. Resetting means stepping back: Are we focused on outcomes or activities? Many have fallen for the Emperor’s New Clothes syndrome and now find themselves as laggards. This trap isn’t just a missed opportunity, it’s a real strategic risk. Competitive advantage is at stake.

As AI integration accelerates, companies will need to balance ‘running the business’ and ‘transforming the business’ simultaneously. How do you architect and then transition your processes for fully agentic workflows?

Summary.The most successful AI startups make focused, grounded bets and resist the urge to spread AI everywhere without specific outcomes in mind. They avoid ‘innovation overload’ that has defined many AI investments: dozens of disconnected pilots that never scale. Ensure your product identifies high-impact, inter-related internal and external use cases and concentrate  efforts there, not fragmented across dozens of micro-tasks.

Experimentation with AI tools sparks innovation and cultural change, all good, but it does not simply fall into enterprise-wide impact. Without clear leadership these efforts remain fragmented, siloed, and ultimately fail. Don’t let your AI product become a vanity project. Companies that act now will turn AI into real results. The rest will be catching up.

A winning AI startup isn’t about chasing buzzwords or deploying models for their own sake, it’s about aligning technology, people, and processes to create durable, compounding advantage.: AI should solve real business problems, not abstract ‘innovation goals’. Ask yourself:

• Where will AI move the needle? With most impact (e.g. revenue uplift, cost reduction, risk mitigation, speed).

• Which use cases matter most? prioritised by value, feasibility, and urgency.

For your startup offering to add value, ypu need to work with organisations that have high-quality, accessible data and a robust AI architecture and platform that is forward leaning and deigned for scalability not just proof of concepts, supported by the right talent and strong governance, ethics and risk controls. From this foundation, AI is most powerful when it creates flywheels, such as

• More usage → more data → better models → more usage

• Automation → lower cost → reinvest savings → better product

• Personalisation → higher engagement → better learning loops

To build your brand with credibility and trust in the fast-moving domain of AI, pilot a few high-value, low-resistance use cases that prove impact quickly, build a reusable playbook, and focus on compounding advantage on what AI can do for your clients. 

Gemini 3.0 is strong evidence that pre-training model scaling still works when algorithmic improvements meet better computing capacity, and Blackwell’s extra power will translate directly into better model capabilities, not just cost efficiency. Together, these two points dismantle the AI bubble murmurings.