Do these 3 things before closing this tab:
1Clear out junk files and repair common Windows errors2Fix the driver behind crashes, sound loss and screen glitches3Repair Windows errors before they cause bigger problemsSome links on this page are affiliate links: if you buy through them we may earn a commission, at no extra cost to you.
An AI startup can reach customers faster than many earlier software companies—but it may depend on a handful of model and cloud providers for the computing, capabilities and distribution that make its product work. That tension was central to investor John Stanton’s November 2023 remarks in Seattle: founders still need a large opportunity, a strong team and capital, but they also need to understand who controls the infrastructure beneath the business and avoid assuming that a major technology company will eventually buy them.
Stanton’s comments are a historical snapshot, not a forecast of 2026 market conditions. They are most useful as a framework for decisions about platform dependence, fundraising and exit options. Since the event, regulators have examined major AI partnerships as well as acquisitions, making the distinction between having a strategic relationship with a platform and remaining independent from it especially important.
![LLC Beginner's Guide [All-in-1]: Everything on How to Start, Run, and Grow Your First Company Without Prior Experience. Includes Essential Tax Hacks, Critical Legal Strategies, and Expert Insights](https://m.media-amazon.com/images/I/41o3X44QPLL._SL160_.jpg)
What John Stanton argued—and when
On November 1, 2023, John Stanton joined Leslie Feinzaig, founder of Graham & Walker, for a Harvard Business School Rock Center “Rock On The Road” event at Pioneer Square Labs in Seattle. Laurie Bishop of the Rock Center helped lead the event. Stanton was speaking as managing director of Trilogy Equity Partners, a longtime business leader and wireless-industry pioneer, a Seattle business figure, and a Microsoft board member at the time. His remarks were a public conversation, not a formal Trilogy investment policy or a statement of Microsoft strategy. GeekWire’s report of the event was published November 6, 2023.
The Tool Desk
Outbyte Driver Updater FREEFix the driver behind crashes, sound loss and screen glitchesFind Drivers →Outbyte PC Repair FREERepair Windows errors before they cause bigger problemsFix Now →Stanton’s basic startup formula was familiar: a disruptive idea, a large market, a capable and diverse team, and enough capital to build the company. What he thought AI changed was the relationship between the startup and the companies that own its models and computing platforms. A startup could build a valuable product without training a frontier model, but it might still need an ongoing commercial or technical relationship with a small number of powerful providers.
#1 Best Overall
- If you want to build a better future, you must believe in secrets.
- The great secret of our time is that there are still uncharted frontiers to explore and new inventions to create. In Zero to One, legendary entrepreneur and investor Peter Thiel shows how we can find singular ways to create those new things.
He also questioned whether founders should rely on the old assumption that a large technology company would be the natural buyer. Regulatory scrutiny could make some acquisitions harder, while large companies might choose to build capabilities internally. Those were Stanton’s assessments in 2023, not a settled prediction about every AI company or transaction.
Capital markets: a fund’s pace is not the whole market
Stanton pushed back on describing the venture market simply as a downturn. He said Trilogy had continued making about one equity investment per quarter during the period he discussed and described venture activity as cyclical, with capital, talent and promising ideas moving in and out of favor.
That account describes one investor’s experience, not total market funding or every founder’s ability to raise on workable terms. A fund can keep investing while the market overall becomes more selective. Likewise, capital can be available but come with lower valuations, more dilution, stronger investor protections or longer diligence. Enthusiasm for AI infrastructure or model companies does not automatically mean that application startups can raise easily.
Recommended Free Tools
For a founder, the useful question is not merely “Is there money for AI?” It is “Is capital available for this company, at a price and with terms that let it reach the next milestone without compromising its ability to operate or raise again?” Funding can be concentrated at the model and infrastructure layers while application companies face pressure to show customer retention, measurable value and credible margins.
Why AI businesses can be more dependent than ordinary software
Stanton contrasted AI startups with some earlier software companies that could get started with relatively limited resources. Wireless networks and cloud platforms, by contrast, required substantial infrastructure investment. Many AI startups sit somewhere between those models: they may not own large data centers or train foundation models, but their products can rely on rented compute, external models and a provider’s terms, capacity and continued service.
That dependence is not automatically a flaw. Using a third-party model can cut development time, avoid the cost of training a large model and let a small team test whether customers will pay for a solution. The question is whether the startup has built something valuable around that capability—or is offering a feature the provider can readily add to its own product.
| Dependency | Questions a founder should answer |
|---|---|
| Foundation model | Can another model perform the essential task at acceptable quality, or would switching require a major rewrite? |
| Cloud and compute | What happens if capacity is constrained, a region becomes unavailable or compute costs rise? |
| API terms and policy | Can pricing, rate limits, acceptable-use rules or model availability change in ways that break the product? |
| Data | Does the company own the data, have a license to use it for this purpose, or merely have temporary access? What do customer and provider contracts permit? |
| Distribution | Does the startup own its customer relationships, or can an app store, cloud provider or model company control access? |
| Economics | Does revenue per task exceed the cost of inference, human review, support and other delivery costs—and does the margin hold as usage grows? |
| Talent | Can the company recruit and retain the people it needs when major platforms can compete for the same specialists? |
| Exit | Can the company continue operating independently, and are there plausible buyers beyond one large platform? |
In January 2025, the Federal Trade Commission described potential competitive concerns in major partnerships involving Microsoft and OpenAI, Amazon and Anthropic, and Google and Anthropic. The FTC staff report put investments across those three partnerships at more than $20 billion and discussed issues including cloud commitments, switching costs, access to business or technical information, and possible effects on access to computing resources and other inputs. The report identified potential concerns; it was not a final finding that each partnership violated antitrust law. Read the FTC’s summary of its AI partnerships and investments study.
The implications reach beyond an acquisition. A startup can remain legally independent yet become commercially dependent on a provider for its capacity, model access or customer channel. Conversely, a partnership can supply useful resources and credibility while creating switching costs or information-sharing risks. The terms—and the ability to keep serving customers if the relationship changes—matter as much as the headline investment.
Build a moat above the model layer
Not every AI company needs to build its own model. Model builders, infrastructure firms, workflow applications, services-enabled software businesses, and data or evaluation companies have different capital needs and sources of defensibility. An application startup may be better off using external models and concentrating its scarce resources on the customer problem.
More durable advantages often come from what surrounds the model:
- Workflow integration: The product fits into a customer’s actual process, permissions and systems rather than producing a compelling standalone demo.
- Distribution and trust: The company can reach buyers, earn adoption and retain a direct customer relationship.
- Legitimate data rights: The company has durable rights to relevant data, and its use complies with contracts, privacy obligations and customer expectations.
- Domain knowledge: The team understands the work, edge cases and consequences well enough to design a product that solves a real operational problem.
- Evaluation and reliability: The company can measure performance on representative tasks, catch failures and improve quality in production.
- Economics: It delivers an outcome customers value at a cost that supports a sustainable margin, including when human review is needed.
- Switching costs rooted in value: Customers would incur real disruption by changing providers because the startup is embedded in a useful workflow—not merely because data is difficult to export.
A useful test is to ask what remains valuable if the underlying model becomes cheaper, more capable or bundled into a platform’s product. If the answer is only “our prompt,” the business may be exposed. If it is trusted distribution, domain-specific workflow, licensed data, reliable implementation and measurable customer results, the startup has a stronger case—though no moat is permanent.
What’s actually slowing this PC down?
Pick the symptom - the matching free tool is one click away.
Manage provider risk without pretending to eliminate it
A model-agnostic design can lower switching costs, but it cannot eliminate dependence on provider pricing, capacity, quality, contract terms or specialized features. Nor is it always sensible to support many models from day one: integrations, testing and operational complexity cost time and money. The goal is to understand and contain the risk that matters to the business.
- Separate product logic from model-specific code where practical. Keep prompts, tool calls and provider-specific features from becoming inseparable from core business logic.
- Benchmark alternatives on real workloads. Compare quality, latency, reliability and total cost on representative customer tasks, not only public demonstrations.
- Track contribution economics. Measure inference and human-review cost by customer and workflow. Identify whether usage growth improves margins or makes losses grow faster.
- Plan a fallback. Decide what the product can do during an outage or capacity shortage: switch providers, queue tasks, offer reduced functionality or notify customers.
- Review contracts and data handling. Understand retention, training use, security commitments, service levels, termination rights, rate limits and what happens to data at the end of the relationship.
- Stress-test price and capacity. Model whether the business survives materially higher inference costs, and know the point at which a workflow must be repriced, redesigned or discontinued.
- Document what is proprietary. Keep a clear account of the company’s own software, data rights, evaluation methods, customer relationships and other assets.
These measures reduce exposure; none guarantees that a provider will preserve a particular model, price or service level. A founder should make that residual risk visible in the business plan and investor discussions.
Raise for a milestone, not for a headline valuation
Capital can buy time, technical talent, distribution, compute access, regulatory expertise or enterprise credibility. It can also impose dilution, governance rights and obligations that constrain later decisions. Founders should therefore compare financing offers on more than the amount raised or the valuation attached to the round.
The Securities and Exchange Commission notes that terms such as “seed,” “Series A” and “Series B” are industry labels, not separate legal categories under federal securities laws. A securities offering generally must be registered or qualify for an exemption; the SEC identifies Rule 506(b) among commonly used private-offering pathways. Its guidance also warns that financing choices and investor terms can affect future fundraising and exits. See the SEC’s guidance on raising later-stage capital. Fundraising rules vary with the offering and jurisdiction; founders should use qualified legal and financial advisers rather than treating a funding label or online template as legal advice.
Rank #4
Before accepting a term sheet or strategic investment, understand at least:
- Dilution: What percentage will founders and employees own after the financing and any option-pool expansion? How might later rounds dilute them?
- Liquidation preferences: Who gets paid first in a sale or liquidation, and how do the preference and any participation rights change what common shareholders receive at different sale prices?
- Control and consent: Which board seats, protective provisions, information rights or other approvals does an investor receive? Could they complicate a sale, a new financing or a change of provider?
- Future compatibility: Are anti-dilution and pro rata provisions workable if the company needs a down round or a longer-than-expected runway?
- Conflicts: Does an investor also finance or advise a competitor, operate a platform the startup depends on, or have access to sensitive information?
- Strategic restrictions: Do exclusivity, customer, data or change-of-control terms affect the company’s ability to work with other providers or buyers?
- Runway and milestones: Does the amount raised give the company time to show evidence of retention, unit economics or product-market fit before it must return to the market?
A strategic investment from a platform provider can bring resources and commercial access. It can also tie the company more closely to that provider or create conflicts with other customers, partners and future investors. A founder should compare the practical benefit with the restrictions, information flows and alternative providers—not assume that a corporate investor will eventually acquire the company.
Independent reader supportYour contribution helps us test, update, and keep practical guides available for everyone.Do not build the plan around one acquirer
Stanton’s 2023 concern was that regulatory scrutiny could make acquisitions by the largest technology companies less dependable, while those companies might build capabilities internally. It helps to separate three questions: Does a buyer want the technology? Does it want the team or customers? Can it complete the transaction on acceptable terms and within a workable timeframe?
The third question is uncertain. The FTC and Department of Justice share merger-review jurisdiction. Depending on the matter, an investigation can be closed, a settlement can be reached, or the government can seek to block a deal in court. Review can add time, expense, uncertainty and potential remedies; it does not mean that regulators categorically prohibit large technology companies from acquiring AI startups. The FTC’s merger-review overview describes the process and possible outcomes. Its premerger notification program says updated Hart-Scott-Rodino forms became effective February 10, 2025; companies considering a transaction need to check current filing requirements and thresholds with counsel, rather than rely on an old rule of thumb.
Quick wins for a faster PC:
Clear out junk files and repair common Windows errorsFree Scan →Scan for outdated or missing drivers - takes under a minuteDriver Scan →Repair Windows errors before they cause bigger problemsFix Now →AI-related acquisitions still occur, but examples do not establish that every startup has an available buyer or that a large-platform deal will clear review. In a February 2026 SEC filing, Grab announced an agreement to acquire an initial 50.1% interest in Stash Financial at an enterprise value of $425 million; the filing described Stash as an AI-powered investing app. That is evidence against saying AI acquisitions have stopped, not evidence that major U.S. platforms can freely acquire any AI business. See the SEC-filed announcement.
Best Value
A more resilient company considers several possible outcomes:
- Acquisition by a mid-sized strategic buyer or a vertical software company.
- Sale to a cloud, cybersecurity, data or enterprise-services provider.
- Growth with private-equity backing or a recapitalization.
- Secondary sales that provide some liquidity to employees or early investors, if permitted and available.
- Merger with a complementary startup.
- Continued independent operation, with an IPO only if scale, revenue quality, governance and market conditions support it.
- As a downside, an asset sale or acqui-hire that may return less value than expected to founders or investors.
“Exit optionality” is shaped years before a sale. Board and investor consent rights, liquidation preferences, customer concentration, data and intellectual-property ownership, exclusivity clauses, provider contracts and the ability to keep operating independently can all affect the choices available. Build a business that can stand on its own; treat a particular acquisition as one possible outcome, not the repayment plan.
Seattle’s capital gap: scarcity can cut both ways
Stanton described Seattle as a place with strong people and ideas but less local capital than founders would like. He saw possible opportunity for regional funds if less investor competition made it easier to find promising companies. Feinzaig also described Seattle founders’ recurring need to seek investment beyond the region. This is an attributed view of Seattle’s market, not proof that the city is universally better for startups or investors than the Bay Area.
Seattle brings deep technology talent, major technology employers and cloud companies, and networks that can help founders recruit or understand enterprise buyers. Less crowded local deal flow may give some funds a chance to identify companies early. But fewer local funds can mean fewer competing offers, less local later-stage capital and a need to build relationships with investors elsewhere. Founders should judge an investor by relevant expertise, follow-on capacity, useful introductions and alignment—not geography alone.
There is also a platform-dependence wrinkle. A region’s strength in cloud and technology employers can help a startup find talent and customers, while concentrating its network around a few large companies that may also be providers, competitors, investors or potential buyers. That makes contract discipline and alternative relationships especially useful, whether a company is based in Seattle or elsewhere.
A founder’s practical decision checklist
- What customer problem is valuable even if a model provider adds similar functionality?
- Which parts of the product, data, evaluation, workflow and distribution are genuinely proprietary or difficult to replicate?
- How many providers can deliver the core capability at acceptable quality, and what would migration cost in time and money?
- What happens to gross margin if inference costs rise, capacity tightens or human review proves more expensive than expected?
- Who owns or licenses each important dataset, and do customer, vendor and privacy terms permit the intended use?
- Can the company keep serving customers during a provider outage, model change or policy shift?
- What does the company need capital to prove before its next raise, and are the terms compatible with that plan?
- Could any investor rights, preferences, exclusivity provisions or change-of-control terms make a future financing or sale harder?
- Can the business survive without being acquired by Google, Microsoft, Amazon or another named platform?
- Which mid-sized, vertical or adjacent-industry buyers might value the product—and can the company operate independently if none appears?
- Can the team cover not only engineering but also product, domain knowledge, security, privacy, legal and regulatory interpretation, sales, implementation and customer support?
Stanton’s thesis remains most useful when read as a set of questions rather than a prediction: Can the company create value beyond rented model capability? Can it finance the work without surrendering more flexibility than it needs to? Can it keep serving customers if a provider or buyer changes course? For AI founders, those are not just technical questions. They are central to the business’s economics, bargaining power and eventual choices.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
Free tools Windows power users keep installed
One-click scans. No signup required.
