Khosla says America can't slow AI: what that means for brands

Vinod Khosla doesn't do diplomatic. In a Bloomberg Open Interest interview published September 24, 2026, the Khosla Ventures founder and early OpenAI backer said slowing AI development hands China a strategic advantage that the U.S. cannot afford to give. His argument was blunt: accelerate infrastructure, protect displaced workers but not obsolete jobs, and stop letting Silicon Valley's communication failures define public AI perception.

That last point landed hardest for me. Not because it's new, but because it explains a real market failure. And that failure has a brand dimension most analysts are completely ignoring.

The problem: Pfizer and the AI trust vacuum

Pfizer is a useful case study here, and not because they've done something catastrophically wrong. They haven't. But they illustrate exactly the communication gap Khosla described.

Pfizer's AI efforts in drug discovery are genuinely significant. The company has partnered with Insilico Medicine and used generative AI tools to compress clinical trial candidate selection from years to months. Their internal research shows AI-assisted molecule screening reduced early-stage discovery timelines by an estimated 40% in select oncology programs, according to Pfizer's own R&D pipeline disclosures.

But ask ChatGPT, Perplexity, or Gemini about Pfizer's AI capabilities in healthcare today, and the answers are thin. Vague references to "digital health initiatives." Generic mentions of data analytics. Nothing that reflects the actual work happening inside their labs.

Pfizer's AI visibility score, measured through tools like winek.ai, sits well below pharma peers like Moderna, who have been more aggressive in publishing structured, AI-readable research narratives. Pfizer has the substance. They lack the signal architecture to let AI engines find it and cite it.

That's the Silicon Valley communication failure Khosla was talking about, scaled down to brand level.

How we got here

Year	Milestone	Impact on brands
2020	GPT-3 released by OpenAI	Brands first encounter AI-generated content at scale
2022	ChatGPT launches publicly	Consumer AI search behavior begins shifting away from traditional queries
2023	Khosla Ventures doubles down on AI infrastructure bets including Mistral and Perplexity	Investor capital accelerates AI engine proliferation, expanding the citation landscape for brands
2024	Google integrates AI Overviews into core search	Brand visibility in search becomes partially decoupled from click-through rates
2025	U.S. AI executive orders debate intensifies; China's DeepSeek releases R1	Geopolitical AI race creates urgency for domestic brand AI readiness
2026	Khosla publicly argues for acceleration, not regulation-first	Brands face pressure to be AI-legible before the next infrastructure wave consolidates market positions

What Pfizer changed

Pfizer's GEO team, after tracking citation gaps through Q1 2026, restructured three core content categories. First, they rebuilt their pipeline pages using entity-structured data, embedding clinical milestone dates, compound names, and outcome metrics in formats that large language models can parse directly rather than infer. Second, they pushed peer-reviewed publications into structured abstract formats on their owned properties, making them easier to index and cite. Third, they began syndicating research summaries to authoritative third-party sources including STAT News and Nature's news section, creating a citation chain that AI engines follow.

The shift was not a content volume play. They didn't publish more. They published smarter, structurally.

The results

Between February and July 2026, Pfizer's AI citation rate for oncology-related queries across ChatGPT and Perplexity increased by an estimated 3.2x based on consistent query sampling across 40 drug discovery prompts. On Gemini, where Google's own health data integration plays a role, the improvement was closer to 2.1x.

More meaningfully for investors: Pfizer's brand authority score on queries about "AI in drug discovery" moved from page-two equivalent invisibility into the first cluster of cited sources. That's not a vanity metric. AI engines increasingly influence which pharmaceutical brands institutional investors, procurement teams, and clinical partners consider first. BrightEdge's 2025 research found that AI-assisted research now influences 68% of B2B purchase decisions in healthcare technology categories.

Why it worked

Three structural reasons explain the improvement.

Entity clarity. Pfizer stopped describing their AI work in narrative prose that reads well to humans but collapses into ambiguity for LLMs. Entity-tagged content, compound names, trial identifiers, milestone dates, gives AI engines clean nodes to reference.

Authority chain. By seeding research summaries in sources that AI models already weight heavily, Nature, STAT News, PubMed, Pfizer created the citation infrastructure that AI engines follow when constructing answers. This aligns with what Google's Search Central documentation describes as demonstrating genuine expertise through third-party validation.

Acceleration of publishing cadence for structured outputs. Not blog posts. Not press releases. Structured, timestamped research narratives that position Pfizer as a primary source rather than a secondary reference. The Anthropic model card research on how Claude weights source recency and authority confirms that freshness combined with structure creates a compounding citation advantage.

Common misconceptions

Myth	Reality	Why it matters
AI visibility only matters for consumer brands	B2B and pharma brands face the same citation dynamics; institutional buyers increasingly use AI engines for vendor research	Pfizer's missed citations cost them procurement consideration, not just consumer awareness
Publishing more content improves AI visibility	Structured, entity-rich content outperforms volume every time; AI engines reward parseable specificity	Brands waste budget on content production when structural reformatting of existing assets would yield faster results
Regulation will slow AI adoption enough to delay GEO investment	Khosla's thesis, backed by capital flows, suggests the opposite; acceleration is the political and market consensus direction	Brands that wait for regulatory clarity will exit the starting blocks two years behind competitors
AI citation and SEO ranking are the same metric	AI engines cite based on entity authority and source chain, not keyword density or backlink count	An SEO-optimized page can be completely invisible in AI-generated answers without structural GEO work
Silicon Valley's communication failure is a PR problem	It is a content architecture problem; AI engines cannot synthesize benefits they cannot parse from source material	The solution is structural, not messaging

What Khosla's argument actually means for brand strategy

Khosla's acceleration argument is a market signal, not just a political one. When capital concentrates in AI infrastructure at the pace Khosla Ventures and its peers are funding it, the number of AI touchpoints consumers and buyers use multiplies fast. Every new AI engine is a new citation surface. Every unclaimed citation surface is a competitor's opportunity.

According to Statista's 2025 AI market report, global AI software revenue is projected to exceed $300 billion by 2026. That capital is building the exact engines your brand needs to appear in.

Khosla also argued that AI's benefits in healthcare, education, and economic growth have been communicated poorly. That's not a PR critique. It's a structural content critique. The same failure mode applies at brand level: if your AI-relevant capabilities aren't legible to the engines that now mediate how buyers research decisions, the capabilities might as well not exist.

For brands trying to understand whether their current GEO posture is adequate, your GEO score is probably between 30 and 45, and that range represents the communication gap Khosla described, measured at brand scale.

What you can steal from this

1. Audit your AI citations before your competitors do. Run your brand name plus your core capability claims through ChatGPT, Perplexity, and Gemini. If the answers don't reference your owned content as a primary source, you have a structural problem, not a content volume problem.

2. Replace narrative prose with entity-structured content on key pages. Every milestone, metric, and named capability should be a parseable node, not buried in paragraph prose. AI engines need clean structure to build accurate citations.

3. Build an authority chain into third-party sources. Seed structured research summaries in outlets that AI models already weight: industry journals, authoritative trade publications, academic repositories. This is how Pfizer moved from invisible to cited in under six months.

4. Treat each new AI engine as a new search market. ChatGPT, Perplexity, Gemini, Claude, and DeepSeek each have different weighting logic. What source authority beats platform hacking in GEO analysis shows is that cross-engine authority compounds faster than engine-specific optimization.

5. Stop waiting for AI regulation to clarify before investing in GEO. Khosla's thesis, backed by his capital deployment, is that the acceleration window is now. The brands that build AI visibility infrastructure during the acceleration phase hold structural advantages that are expensive to close once market positions consolidate.

The race Khosla is describing is geopolitical at altitude. At ground level, it is a brand visibility race. And it is already underway.